Sometimes it is useful to “backcast” a time series — that is, forecast in reverse time. Although there are no in-built R functions to do this, it is very easy to implement. Suppose x is our time series and we want to backcast for periods. Here is some code that should work for most univariate time series. The example is non-seasonal, but the code will also work with seasonal data. library(forecast) x <- WWWusage h <- 20 f <- frequency(x) # Reverse time revx <- ts(rev(x), frequency=f) # Forecast fc <- forecast(auto.arima(revx), h) plot(fc) # Reverse time again fc$mean <- ts(rev(fc$mean),end=tsp(x)[1] - 1/f, frequency=f) fc$upper <- fc$upper[h:1,] fc$lower <- fc$lower[h:1,] fc$x <- x # Plot result plot(fc, xlim=c(tsp(x)[1]-h/f, tsp(x)[2]))

So far, we have just put the data in and out. And we have had a pretty good track record doing so. However, what do we do with the data now that we have it? As you can expect, we need to read it out. Usually by specific date ranges. The interesting thing is that we usually are not interested in just a single channel, we care about multiple channels. And for fun, those channel might be synchronized or not. An example of the first might be the current speed and the current engine temperature in a car. They are generally share the exact same timestamps. An example of out of sync is when you have a sensor on a rooftop measuring rainfall, and another sensor in the sewer measuring water flow rates. (Again, thanks to Dan for helping me with the domain). This is interesting, because it present quite a few interesting problems: We need to merge different streams into a unified view. We need to handle both matching and non matching sequences. We need to handle erroneous data, what happens when we have two reading for the same time for the same sensor? Yes, that shouldn’t happen, but it does. I solved this with the following API: public class RangeEntry { public DateTime Timestamp; public double?[] Values; } IEnumerable results = dts.ScanRanges(DateTime.MinValue, DateTime.MaxValue, new[] { "6febe146-e893-4f64-89f8-527f2dbaae9b", "707dcb42-c551-4f1a-9203-e4b0852516cf", "74d5bee8-9a7b-4d4e-bd85-5f92dfc22edb", "7ae29feb-6178-4930-bc38-a90adf99cfd3", }); This API gives me the results in the time order, with the same positions as the ids requested for the values. With nulls if there isn’t a value matching the value from that time in that particular sensor channel. The actual implementation relies on this method: IEnumerable ScanRange(DateTime start, DateTime end, string id) All this does it provide the entries all the entries in a particular date range, for a particular channel. Let us see how we implement multi channel scanning on top of this: private class PendingEnumerator { public IEnumerator Enumerator; public int Index; } private class PendingEnumerators { private readonly SortedDictionary> _values = new SortedDictionary>(); public void Enqueue(PendingEnumerator entry) { List list; var dateTime = entry.Enumerator.Current.Timestamp; if (_values.TryGetValue(dateTime, out list) == false) { _values.Add(dateTime, list = new List()); } list.Add(entry); } public bool IsEmpty { get { return _values.Count == 0; } } public List Dequeue() { if (_values.Count == 0) return new List(); var kvp = _values.First(); _values.Remove(kvp.Key); return kvp.Value; } } public IEnumerable ScanRanges(DateTime start, DateTime end, string[] ids) { if (ids == null || ids.Length == 0) yield break; var pending = new PendingEnumerators(); for (int i = 0; i < ids.Length; i++) { var enumerator = ScanRange(start, end, ids[i]).GetEnumerator(); if(enumerator.MoveNext() == false) continue; pending.Enqueue(new PendingEnumerator { Enumerator = enumerator, Index = i }); } var result = new RangeEntry { Values = new double?[ids.Length] }; while (pending.IsEmpty == false) { Array.Clear(result.Values,0,result.Values.Length); var entries = pending.Dequeue(); if (entries.Count == 0) break; foreach (var entry in entries) { var current = entry.Enumerator.Current; result.Timestamp = current.Timestamp; result.Values[entry.Index] = current.Value; if(entry.Enumerator.MoveNext()) pending.Enqueue(entry); } yield return result; } } We are getting a single entry from each channel into the pending enumerators. Then, we collate all the entries that share the same time into a single entry. We use the Index property to track the actual expected index of the entry in the output. And we handle duplicate times in the same channel by outputting multiple entries. Testing this on my 1.1 million records data set, we can get 185 thousands records back in 0.15 seconds.

The best code coverage metric for Scala is statement coverage. Simple as that. It suits the typical programming style in Scala best. Scala is a chameleon and it can look like anything you wish, but very often more statements are written on a single line and conditional “if” statements are used rarely. In other words, line coverage and branch coverage metrics are not helpful. Java tools Scala runs on JVM and therefore many existing tools for Java can be used for Scala as well. But for code coverage it’s a mistake to do so. One wrong option is to use tools that measure coverage looking at bytecode like JaCoCo. Even though it gives you a coverage rate number, JaCoCo knows nothing about Scala and therefore it doesn’t tell you which piece of code you forgot to cover. Another misfortune are tools that natively support line and branch coverage metrics only. Cobertura is a standard in Java world and XML coverage report that it generates is supported by many tools. Some Scala code coverage tools decided to use Cobertura XML report format because of its popularity. Sadly, it doesn’t support statement coverage. Statement coverage Why? Because a typical Scala statement looks like this (a single line of code): def f(l: List[Int]) = l.filter(_ > 0).filter(_ < 42).takeWhile(_ != 3).foreach(println(_)) Neither line nor branch coverage works here. When would you consider this single line as being covered by a test? If at least one statement of that line has been called? Maybe. Or all of them? Also maybe. Where is a branch? Yes, there are statements that are executed conditionally, but the decision logic is hidden in internal implementation of List. Branch coverage tools are helpless, because they don't see this kind of conditional execution. What we need to know instead is whether individual statements like _ > 0, _ < 42 or println(_) have beed executed by an automated test. This is the statement coverage. Scoverage to the rescue! Luckily there is a tool named Scoverage. It is a plugin for Scala compiler. There is also a plugin for SBT. It does exactly what we need. It generates HTML report and also own XML report containing detailed information about covered statements. Scoverage plugin for SonarQube Recently I have implemented a plugin for Sonar 4 so that statement coverage measurement can become an integral part of your team's continuous integration process and a required quality standard. It allows you to review overall project statement coverage as well as dig deeper into sub-modules, directories and source code files to see uncovered statements. Project dashboard with Scoverage plugin: Multi-module project overview: Columns with statement coverage, total number of statements and number of covered statements: Source code markup with covered and uncovered lines:

This morning an interesting topic was posted to the Apache ServiceMix user forum, asking the question: To ServiceMix or not ServiceMix. In my mind the short answer is: NO Guillaume Nodet one of the key architects and long time committer on Apache ServiceMix already had his mind set 3 years ago when he wrong this blog post - Thoughts about ServiceMix. What has happened on the ServiceMix project was that the ServiceMix kernel was pulled out of ServiceMix into its own project - Apache Karaf. That happened in spring 2009, which Guillaume also blogged about. So is all that bad? No its IMHO all great. In fact having the kernel as a separate project, and Camel and CXF as the integration and WS/RS frameworks, would allow the ServiceMix team to focus on building the ESB that truly had value-add. But that did not happen. ServiceMix did not create a cross product security model, web console, audit and trace tooling, clustering, governance, service registry, and much more that people were looking for in an ESB (or related to a SOA suite). There were only small pieces of it, but never really baked well into the project. That said its not too late. I think the ServiceMix project is dying, but if a lot of people in the community step up, and contribute and work on these things, then it can bring value to some users. But I seriously doubt this will happen. PS: 6 years ago I was working as a consultant and looked at the next integration platform for a major Danish organization, and we looked at ServiceMix back then and dismissed it due its JBI nature, and the new OSGi based architecture was only just started. And frankly it has taken a long long time to mature Apache Karaf / Felix / Aries and the other pieces in OSGi to what they are today to offer a stable and sound platform for users to build their integration applications. That was not the case 4-6 years ago. Okay No to ServiceMix - what are my options then? So what should use you instead of ServiceMix? Well in my mind you have at least these two options. 1) Use Apache Karaf and add the pieces you need, such as Camel, CXF, ActiveMQ and build your own ESB. These individual projects have regular releases, and you can upgrade as you need. The ServiceMix project only has the JBI components in additional, that you should NOT use. Only legacy users that got on the old ServiceMix 3.x wagon may need to use this in a graceful upgrade from JBI to Karaf based containers. 2) Take a look at fabric8. IMHO fabric8 is all that value-add the ServiceMix project did not create, and a lot more. James Strachan, just blogged today about some of his thoughts on fabric8, JBoss Fuse, and Karaf. I encourage you to take a read. For example he talks about how fabric becomes poly container, so you have a much wider choice of which containers/JVM to run your integration applications. OSGi is no longer a requirement. (IMHO that is very very existing and potentially a changer). I encourage you to check out fabric8 web-site, and also read the overview and motivation sections of the documentation. And then check out some of the videos. After the upcoming JBoss Fuse 6.1 release, the Fuse team at Red Hat will have more time and focus to bring the documentation at fabric8 up to date covering all the functionality we have (there is a lot more), and as well bring out a 1.0 community released using pure community releases. This gives end users a 100% free to use out of the box release. And users looking for a commercial release can then use JBoss Fuse. Best of both worlds. Summary Okay back to the question - to ServiceMix or not. Then NO. Innovation happens outside ServiceMix, and also more and more outside Apache. If you have thoughts then you can share those in comments to this blog, or better yet, get involved in the discussion forum at the ServiceMix user forum. PPS: The thoughts on this blog is mine alone, and are not any official words from my employer.

Background SPNego is RFC 4178 used for negotiation either NTLM or Kerberos based SSO. A typical use case is for web applications to reuse the authentication used by Desktops such as Windows or Linux. In this article, we will explore approaches for SPNego authentication with JBoss Enterprise Application Platform. JBoss Negotiation is the library that provides the SPNego authentication support in JBoss. This library has been integrated in JBoss EAP and WildFly Application Server. Checklist Obtain JBoss EAP from jboss.org. Enable your JavaEE Web Application for SPNego Authentication. Configure JBoss EAP for SPNego. Configure your Browsers for SPNego. Start JBoss EAP. Test your web application. Obtain JBoss EAP from jboss.org Download JBoss EAP 6.2 or newer from http://www.jboss.org/products/eap You can also use WildFly Application Server from http://www.wildfly.org. Your configuration may vary slightly. Enable your JavaEE Web Application for SPNego Authentication It is easier to use a demo web application as a starting point. You can then modify your web application for SPNego authentication. The demo web application we use for this article is called spnego-demo, by my colleague, Josef Cazek. The demo web application is available at https://github.com/kwart/spnego-demo . You can also download the spnego-demo.war from here . Fully configured web application spnego-demo.war can be obtained from this location . Copy the spnego-demo.war in your jboss-eap-6.2/standalone/deployments directory. Configure JBoss EAP for SPNego Authentication You will need to configure a couple of security domains and system properties in JBoss EAP6. There are two ways by which you can configure: either manual editing or using CLI tool. Manual Editing of configuration file standalone.xml in jboss-eap-6.2/standalone/configuration Add system properties to this file. Remember to put this block right after the extensions block (around line 25 of the configuration file). Add security domains to this file. Remember to put these blocks in the block. Using Command Line Interface to update JBoss EAP Go to the bin directory of JBoss EAP 6.2 and run the following. $ cat << EOT > $SPNEGO_TEST_DIR/cli-commands.txt /subsystem=security/security-domain=host:add(cache-type=default) /subsystem=security/security-domain=host/authentication=classic:add(login-modules=[{"code"=>"Kerberos", "flag"=>"required", "module-options"=>[ ("debug"=>"true"),("storeKey"=>"true"),("refreshKrb5Config"=>"true"),("useKeyTab"=>"true"),("doNotPrompt"=>"true"),("keyTab"=>"$SPNEGO_TEST_DIR/http.keytab"),("principal"=>"HTTP/localhost@JBOSS.ORG")]}]) {allow-resource-service-restart=true} /subsystem=security/security-domain=SPNEGO:add(cache-type=default) /subsystem=security/security-domain=SPNEGO/authentication=classic:add(login-modules=[{"code"=>"SPNEGO", "flag"=>"required", "module-options"=>[("serverSecurityDomain"=>"host")]}]) {allow-resource-service-restart=true} /subsystem=security/security-domain=SPNEGO/mapping=classic:add(mapping-modules=[{"code"=>"SimpleRoles", "type"=>"role", "module-options"=>[("jduke@JBOSS.ORG"=>"Admin"),("hnelson@JBOSS.ORG"=>"User")]}]) {allow-resource-service-restart=true} /system-property=java.security.krb5.conf:add(value="$SPNEGO_TEST_DIR/krb5.conf") /system-property=java.security.krb5.debug:add(value=true) /system-property=jboss.security.disable.secdomain.option:add(value=true) :reload() EOT $ ./jboss-cli.sh -c --file=$SPNEGO_TEST_DIR/cli-commands.txt This is explained in https://github.com/kwart/spnego-demo/blob/master/README.md We will need a keytab file. In this example, we will use the Kerberos Server using ApacheDS (as explained in Appendix A). $ java -classpath kerberos-using-apacheds.jar org.jboss.test.kerberos.CreateKeytab HTTP/localhost@JBOSS.ORG httppwd http.keytab Note that the http.keytab has been configured in the security domain called "host" in standalone.conf. So place the keytab file appropriately while correcting the path defined in security domain. More information is available at https://github.com/kwart/kerberos-using-apacheds/blob/master/README.md JBoss EAP will need a keytab file. In this example we use a keytab called as http.keytab Different tools such as ktutil exist to create keytab files. Keytab files contain Kerberos Principals and encrypted keys. It is important to safeguard keytab files. It is very important that JBoss EAP configuration for keytab in the security domain "host" refers to the actual path of the keytab file. Configure your Browsers for SPNego The browsers such as Microsoft IE, Mozilla Firefox, Google Chrome, Apple Safari have different settings for enabling SPNego or Integrated Authentication. Start JBoss EAP Go to the bin directory of JBoss EAP 6.2 and either use standalone.sh (Unix/Linux) or standalone.bat to start your instance. Test your Web Application Assuming you have followed Appendix A steps to start the kerberos server and done kinit, you are ready to test the web application. In this article we have used spnego-demo, we can test that by going to http://localhost:8080/spnego-demo/ You can click on the "User Page" link and you should be able to see the principal name as "hnelson@jboss.org" Appendix A Local Kerberos Server Download the zip file https://github.com/kwart/kerberos-using-apacheds/archive/master.zip Unzip the zip file into a directory. Build the package using maven. $ mvn clean package Start the Kerberos Server as $ java -jar target/kerberos-using-apacheds.jar test.ldif A krb5.conf file has been created. Login now using hnelson@jboss.org $ kinit hnelson@JBOSS.ORG Password for hnelson@JBOSS.ORG: secret Launch Firefox via command line from where the kinit was run On MacOSX $open -a firefox http://localhost:8080/spnego-demo/ Appendix B Kerberos Command Line Utilities klist can be used to see the current kerberos tickets. $ klist Credentials cache: API:501:10 Principal: hnelson@JBOSS.ORG Issued Expires Principal Feb 9 21:19:30 2014 Feb 10 07:19:27 2014 krbtgt/JBOSS.ORG@JBOSS.ORG kdestroy can be used to clear the current kerberos tickets. References SPNego Demo Web Application: https://github.com/kwart/spnego-demo Kerberos Server using ApacheDS: https://github.com/kwart/kerberos-using-apacheds JBoss EAP 6 http://www.jboss.org/products/eap PicketLink Open Source Project: http://www.picketlink.org Troubleshooting https://docs.jboss.org/author/display/PLINK/SPNego+Support+Questions Remember krb5.conf is important for client side kerberos interactions. You can use a environment variable on Unix/Linux/Mac systems called KRB5_CONFIG to point to your krb5.conf Acknowledgement Darran Lofthouse for the wonderful JBoss Negotiation Project and Josef Czacek for the SPNego-demo and Kerberos_using_Apache DS projects.

dan liebster has been kind enough to send me a real world time series database. the data has been sanitized to remove identifying issues, but this is actually real world data, so we can learn a lot more about this. this is what this looks like: the first thing that i did was take the code in this post , and try it out for size. i wrote the following: int i = 0; using (var parser = new textfieldparser(@"c:\users\ayende\downloads\timeseries.csv")) { parser.hasfieldsenclosedinquotes = true; parser.delimiters = new[] {","}; parser.readline();//ignore headers var startnew = stopwatch.startnew(); while (parser.endofdata == false) { var fields = parser.readfields(); debug.assert(fields != null); dts.add(fields[1], datetime.parseexact(fields[2], "o", cultureinfo.invariantculture), double.parse(fields[3])); i++; if (i == 25*1000) { break; } if (i%1000 == 0) console.write("\r{0,15:#,#} ", i); } console.writeline(); console.writeline(startnew.elapsed); } note that we are using a separate transaction per line , which means that we are really doing a lot of extra work. but this simulate very well incoming events coming one at a time. we were able to process 25,000 events in 8.3 seconds. at a rate of just over 3 events per millisecond . now, note that we have in here the notion of “channels”. from my investigation, it seems clear that some form of separation is actually very common in time series data. we are usually talking about sensors or some such, and we want to track data across different sensors over time. and there is little if any call for working over multiple sensors / channels at the same time. because of that, i made a relatively minor change in voron, that allows it to have an infinite number of separate trees. that means that i can use as many trees as you want, and we can model a channel as a tree in voron. i also changed things so we instead of doing a single transaction per line, we will do a transaction per 1000 lines. that dropped the time to insert 25,000 lines to 0.8 seconds. or a full order of magnitude faster. that done, i inserted the full data set, which is just over 1,096,384 records. that took 36 seconds. in the data set i have, there are 35 channels. i just tried, and reading all the entries in a channel with 35,411 events takes 0.01 seconds. that allows doing things like doing averages over time, comparing data, etc. you can see the code implementing this in the following link .

Since JBoss EAP 6.1 / AS 7.2.0 is modular and you can exclude what modules are visible to your webapp, you would expect it to be easy to ignore the built-in implementation of JAX-RS (Rest Easy 2.3.6) and use a custom one (3.0.6). However, sadly, this is not the case. You are stuck with what the official guide suggests, i.e.upgrading Rest Easy globally – provided that no other webapp running on the server becomes broken by the upgrade. This should be enough to exclude the built-in Rest Easy and be able to use a version included in the webapp: However it is far from working. This nearly does the job (though few of the exclusions might be unnecessary): However, only nearly. The problem is that the exclusion of javax.ws.rs.api has no effect. It seems as the core Java EE APIs cannot be excluded. Dead end. BTW, this are my final jax-rs related dependencies: // resteasyVersion = '3.0.6.Final' compile group: 'org.jboss.resteasy', name: 'jaxrs-api', version: resteasyVersion compile group: 'org.jboss.resteasy', name: 'resteasy-jaxrs', version: resteasyVersion compile group: 'org.jboss.resteasy', name: 'resteasy-jackson2-provider', version: resteasyVersion // JSONP compile group: 'org.jboss.resteasy', name: 'async-http-servlet-3.0', version: resteasyVersion // Required at runtime compile group: 'org.jboss.resteasy', name: 'resteasy-servlet-initializer', version: resteasyVersion // Required at runtime An approximate history of failed attempts I do not remember anymore exactly all the dead ends I went through but here is an approximate overview of the exceptions I got at deployment or runtime. java.lang.ClassNotFoundException: org.jboss.resteasy.plugins.server.servlet.HttpServlet30Dispatcher - fixed likely by adding org.jboss.resteasy:async-http-servlet-3.0:3.0.6.Final to the dependencies java.lang.ClassCastException: myapp.rs.RestApplication cannot be cast to javax.servlet.Servlet - fixed likely by adding org.jboss.resteasy:resteasy-servlet-initializer:3.0.6.Final to the dependencies java.lang.NoSuchMethodError: org.jboss.resteasy.spi.ResteasyProviderFactory.(Lorg/jboss/resteasy/spi/ResteasyProviderFactory;)V - fixed likely by adding more of the RestEasy/Jackson modules to the exclusion list java.lang.NoSuchMethodError: org.jboss.resteasy.specimpl.BuiltResponse.getHeaders()Ljavax/ws/rs/core/MultivaluedMap; - this is the ultimate one that cannot be fixed; the problem is that BuiltResponse from resteasy-jaxrs inherits from javax.ws.rs.core.Response however the version of this class from jaxrs-api-3.0.6.Final.jar is ignored in favour of Response from JAX-RS 1.1 from the javax.ws.rs.api module (/jboss-eap-6.1.0/modules/system/layers/base/javax/ws/rs/api/main/jboss-jaxrs-api_1.1_spec-1.0.1.Final-redhat-2.jar), which lacks the getHeaders method and, as mentioned, cannot be excluded. (Thanks to allprog for hinting at this confilct!) Conclusion The only way to use a newer JAX-RS is to upgrade the JBoss modules. If that would break some other webapps, you are stuck. Lessons learned: Application servers with the plenty of out-of-the-box, well-integrated (?) functionality seem attractive but when you run into conflicting libraries and classloading issues, their value diminishes rapidly. Starting with something simple that you control fully, such as Jettty, is perhaps in the long run a better solution. Also, running multiple webapps on the same server was perhaps smart in 2000 but is not worth the pain nowadays. We have plenty of disk space and memory so reuse of libraries is unimportant and the ability to manage global settings for all apps at one place has certainly better alternatives. Microservices FTW! Update: As Yannick has pointed out , the conclusion seems too general and unjustified. That is because I have arrived to it already before and this problem with JBoss serves only as another confirmation. Solution? Bill Burke has proposed a solution : I’ve lived through your pain and here’s a solution that works on AS7.1.1, EAP6.x, and Wildfly: https://github.com/keycloak/keycloak/blob/master/server/src/main/webapp/WEB-INF/jboss-deployment-structure.xml JBoss Modules don’t suck. The implicit dependencies do. The culprit is the “javaee.api” module which you have missing from your exclude. This module includes every single Java EE API. I haven’t tried, but I think if you reduce your excludes to just that module and the “resteasy” subsystem, it will work. ... FYI, I believe the “javaee.api” module problem is fixed in Wildfly so you won’t have to do the extra exclude.

In this post we will see how to setup a load balanced JBoss or Tomcat environment with the fail-over capability. This post assumes that both Jboss/Tomcat and Apache httpd are setup and running properly. Configure Apache Httpd Step 1: Configure apache’s workers.properties Go to /usr/local/apache2/conf/extra and open workers.properties and configure the properties indicated in bold. # for mapping requests # The configuration directives are valid # for the mod_jk version 1.2.18 and later # worker.list=loadbalancer,status # Define node # modify the host as your host IP or DNS name. worker.node.port=8009 worker.node.host=192.168.0.3 #(IP or DNS name of the server on which Jboss is running) worker.node.type=ajp13 worker.node.lbfactor=1 worker.node2.port=8009 worker.node2.host=192.168.0.4 #(IP or DNS name of the server on which Jboss is running) worker.node2.type=ajp13 worker.node2.lbfactor=1 # Load-balancing behaviour worker.loadbalancer.type=lb worker.loadbalancer.balance_workers=node,node2 worker.loadbalancer.sticky_session=1 worker.status.type=status The property balance_workers is used to specify the nodes to load balance. For example, if we specify only ‘node’, all requests will be routed to the server named ‘node’. Modify this to test that requests are going to both servers. Step 2: Configure mod-jk.conf Go to /usr/local/apache2/conf/extra and open mod-jk.conf and configure the bold properties according to your requirements. #Load mod_jk module # Specify the filename of the mod_jk lib LoadModule jk_module modules/mod_jk.so # Where to find workers.properties JkWorkersFile conf/extra/workers.properties # Where to put jk logs JkLogFile /var/apache/logs/mod_jk.log # Set the jk log level [debug/error/info] JkLogLevel debug # Select the log format JkLogStampFormat "[%a %b %d %H:%M:%S %Y]" # JkOptions indicates to send SSK KEY SIZE # Note: Changed from +ForwardURICompat. # See http://tomcat.apache.org/security-jk.html JkOptions +ForwardKeySize +ForwardURICompatUnparsed -ForwardDirectories JkOptions +FlushPackets # JkRequestLogFormat JkRequestLogFormat "%w %V %T" # Mount your applications JkMount /* loadbalancer # You can use external file for mount points. # It will be checked for updates each 60 seconds. # The format of the file is: /url=worker # /examples/*=loadbalancer JkMountFile conf/extra/uriworkermap.properties # Add shared memory. # This directive is present with 1.2.10 and # later versions of mod_jk, and is needed for # for load balancing to work properly # Note: Replaced JkShmFile logs/jk.shm due to SELinux issues. Refer to # https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=225452 JkShmFile /var/run/jk.shm # Add jkstatus for managing runtime data JkMount status Order deny,allow Deny from all Allow from 127.0.0.1 The JkMount /* loadbalancer indicates that all requests are to be routed through the load balancer. Step 3: Go to /usr/local/apache2/conf and edit httpd.conf Add Include conf/extra/mod-jk.conf to the end of the file. Configure Tomcat/JBoss These settings need to be made for each of the nodes we intend on placing behind the load balancer. We need to modify the server.xml file which is located at For Tomcat ../apache-tomcat/conf/ For JBoss ../jboss-5.1.0.GA/server/default/deploy/jbossweb.sar Edit the following tag: where the jvmRoute attribute is configured to ‘node’ (which is the name we gave the first worker while configuring workers.properties.) Next, locate the following tag and edit the port to be the same as while configuring the workers.properties. Configure the other node(s) accordingly. Next, Start Apache httpd and the Jboss/tomcat node(s) that you have configured.

At the end of last year I was playing around with running scheduled tasks to monitor a Neo4j cluster and one of the problems I ran into was that the monitoring would sometimes exit. I eventually realised that this was because a RuntimeException was being thrown inside the Runnable method and I wasn’t handling it. The following code demonstrates the problem: import java.util.ArrayList; import java.util.List; import java.util.concurrent.*; public class RunnableBlog { public static void main(String[] args) throws ExecutionException, InterruptedException { ScheduledExecutorService executor = Executors.newSingleThreadScheduledExecutor(); executor.scheduleAtFixedRate(new Runnable() { @Override public void run() { System.out.println(Thread.currentThread().getName() + " -> " + System.currentTimeMillis()); throw new RuntimeException("game over"); } }, 0, 1000, TimeUnit.MILLISECONDS).get(); System.out.println("exit"); executor.shutdown(); } } If we run that code we’ll see the RuntimeException but the executor won’t exit because the thread died without informing it: Exception in thread "main" pool-1-thread-1 -> 1391212558074 java.util.concurrent.ExecutionException: java.lang.RuntimeException: game over at java.util.concurrent.FutureTask$Sync.innerGet(FutureTask.java:252) at java.util.concurrent.FutureTask.get(FutureTask.java:111) at RunnableBlog.main(RunnableBlog.java:11) at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method) at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:57) at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43) at java.lang.reflect.Method.invoke(Method.java:601) at com.intellij.rt.execution.application.AppMain.main(AppMain.java:120) Caused by: java.lang.RuntimeException: game over at RunnableBlog$1.run(RunnableBlog.java:16) at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:471) at java.util.concurrent.FutureTask$Sync.innerRunAndReset(FutureTask.java:351) at java.util.concurrent.FutureTask.runAndReset(FutureTask.java:178) at java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.access$301(ScheduledThreadPoolExecutor.java:178) at java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.run(ScheduledThreadPoolExecutor.java:293) at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1110) at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:603) at java.lang.Thread.run(Thread.java:722) At the time I ended up adding a try catch block and printing the exception like so: public class RunnableBlog { public static void main(String[] args) throws ExecutionException, InterruptedException { ScheduledExecutorService executor = Executors.newSingleThreadScheduledExecutor(); executor.scheduleAtFixedRate(new Runnable() { @Override public void run() { try { System.out.println(Thread.currentThread().getName() + " -> " + System.currentTimeMillis()); throw new RuntimeException("game over"); } catch (RuntimeException e) { e.printStackTrace(); } } }, 0, 1000, TimeUnit.MILLISECONDS).get(); System.out.println("exit"); executor.shutdown(); } } This allows the exception to be recognised and as far as I can tell means that the thread executing the Runnable doesn’t die. java.lang.RuntimeException: game over pool-1-thread-1 -> 1391212651955 at RunnableBlog$1.run(RunnableBlog.java:16) at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:471) at java.util.concurrent.FutureTask$Sync.innerRunAndReset(FutureTask.java:351) at java.util.concurrent.FutureTask.runAndReset(FutureTask.java:178) at java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.access$301(ScheduledThreadPoolExecutor.java:178) at java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.run(ScheduledThreadPoolExecutor.java:293) at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1110) at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:603) at java.lang.Thread.run(Thread.java:722) pool-1-thread-1 -> 1391212652956 java.lang.RuntimeException: game over at RunnableBlog$1.run(RunnableBlog.java:16) at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:471) at java.util.concurrent.FutureTask$Sync.innerRunAndReset(FutureTask.java:351) at java.util.concurrent.FutureTask.runAndReset(FutureTask.java:178) at java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.access$301(ScheduledThreadPoolExecutor.java:178) at java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.run(ScheduledThreadPoolExecutor.java:293) at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1110) at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:603) at java.lang.Thread.run(Thread.java:722) pool-1-thread-1 -> 1391212653955 java.lang.RuntimeException: game over at RunnableBlog$1.run(RunnableBlog.java:16) at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:471) at java.util.concurrent.FutureTask$Sync.innerRunAndReset(FutureTask.java:351) at java.util.concurrent.FutureTask.runAndReset(FutureTask.java:178) at java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.access$301(ScheduledThreadPoolExecutor.java:178) at java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.run(ScheduledThreadPoolExecutor.java:293) at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1110) at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:603) at java.lang.Thread.run(Thread.java:722) This worked well and allowed me to keep monitoring the cluster. However, I recently started reading ‘Java Concurrency in Practice‘ (only 6 years after I bought it!) and realised that this might not be the proper way of handling the RuntimeException. public class RunnableBlog { public static void main(String[] args) throws ExecutionException, InterruptedException { ScheduledExecutorService executor = Executors.newSingleThreadScheduledExecutor(); executor.scheduleAtFixedRate(new Runnable() { @Override public void run() { try { System.out.println(Thread.currentThread().getName() + " -> " + System.currentTimeMillis()); throw new RuntimeException("game over"); } catch (RuntimeException e) { Thread t = Thread.currentThread(); t.getUncaughtExceptionHandler().uncaughtException(t, e); } } }, 0, 1000, TimeUnit.MILLISECONDS).get(); System.out.println("exit"); executor.shutdown(); } } I don’t see much difference between the two approaches so it’d be great if someone could explain to me why this approach is better than my previous one of catching the exception and printing the stack trace.

I'm passionate about front end optimization and have been for years. My original inspiration was Steve Souders and his Even Faster Web Sites talk at OSCON 2008. Since then, I've optimized this blog, made it even faster with a new design, doubled the speed of several apps for clients and showed how to make AppFuse faster. As part of my Devoxx 2013 presentation, I showed how to do page speed optimization in a Java webapp. I developed a couple AngularJS apps last year. To concat and minify their stylesheets and scripts, I used mechanisms that already existed in the projects. On one project, it was Ant and its concat task. On the other, it was part of a Grails application, so I used the resources and yui-minify-resources plugins. The Angular project I'm working on now will be published on a web server, as well as bundled in an iOS native app. Therefore, I turned to Grunt to do the optimization this time. I found it to be quite simple, once I figured out how to make it work with Angular. Based on my findings, I submitted a pull request to add Grunt to angular-seed. Below are the steps I used to add Grunt to my Angular project. Install Grunt's command line interface with "sudo npm install -g grunt-cli". Edit package.json to include a version number (e.g. "version": "1.0.0"). Add Grunt plugins in package.json to do concat/minify/asset versioning: "grunt": "~0.4.1", "grunt-contrib-concat": "~0.3.0", "grunt-contrib-uglify": "~0.2.7", "grunt-contrib-cssmin": "~0.7.0", "grunt-usemin": "~2.0.2", "grunt-contrib-copy": "~0.5.0", "grunt-rev": "~0.1.0", "grunt-contrib-clean": "~0.5.0" Create a Gruntfile.js that runs all the plugins. module.exports = function (grunt) { grunt.initConfig({ pkg: grunt.file.readJSON('package.json'), clean: ["dist", '.tmp'], copy: { main: { expand: true, cwd: 'app/', src: ['**', '!js/**', '!lib/**', '!**/*.css'], dest: 'dist/' }, shims: { expand: true, cwd: 'app/lib/webshim/shims', src: ['**'], dest: 'dist/js/shims' } }, rev: { files: { src: ['dist/**/*.{js,css}', '!dist/js/shims/**'] } }, useminPrepare: { html: 'app/index.html' }, usemin: { html: ['dist/index.html'] }, uglify: { options: { report: 'min', mangle: false } } }); grunt.loadNpmTasks('grunt-contrib-clean'); grunt.loadNpmTasks('grunt-contrib-copy'); grunt.loadNpmTasks('grunt-contrib-concat'); grunt.loadNpmTasks('grunt-contrib-cssmin'); grunt.loadNpmTasks('grunt-contrib-uglify'); grunt.loadNpmTasks('grunt-rev'); grunt.loadNpmTasks('grunt-usemin'); // Tell Grunt what to do when we type "grunt" into the terminal grunt.registerTask('default', [ 'copy', 'useminPrepare', 'concat', 'uglify', 'cssmin', 'rev', 'usemin' ]); }; Add comments to app/index.html so usemin knows what files to process. The comments are the important part, your files will likely be different. ... A couple of things to note: 1) the copy task copies the "shims" directory from Webshims lib because it loads files dynamically and 2) setting "mangle: false" on the uglify task is necessary for Angular's dependency injection to work. I tried to use grunt-ngmin with uglify and had no luck. After making these changes, I'm able to run "grunt" and get an optimized version of my app in the "dist" folder of my project. For development, I continue to run the app from my "app" folder, so I don't currently have a need for watching and processing assets on-the-fly. That could change if I start using LESS or CoffeeScript. The results speak for themselves: from 27 requests to 5 on initial load, and only 3 requests for less than 2K after that. YSlow Page Speed No optimization 75 27 HTTP requests / 464K 55/100 Apache optimization (gzip and expires headers) 89 initial load: 26 requests / 166K primed cache: 4 requests / 40K 88/100 Apache + concat/minified/versioned files 98 initial load: 5 requests / 136K primed cache: 3 requests / 1.4K 93/100

Introduction Some time ago we decided to upgrade our application from JBoss 5 to 7 (technically 7.2). In this article I going to describe several things which we found problematic. At the end I also provided a short list of benefits we gained in retrospect. First some general information about our application. It was built using EJB 3.0 technology. We have 2 interfaces for communicating with other components – JMS and JAX-WS. We use JBoss AS 5 as our messaging broker which is started as a separate JVM process. This part of the system we were not allowed to change. Finally – we use JPA to store processing results to Oracle DB. Step #1 – Convince your Product Owner Although our application was rather small and built on JEE5 standard it took us 4 weeks to migrate it to JEE6 and JBoss 7. So you can't do it as a maintenance ticket – it's simply too big. There is always problem with providing Business Value of such migration for Product Owners as well as for key Stakeholders. There are several aspects which might help you convincing them. One of the biggest benefits is processing time. JBoss 7 is simply faster and has better caching (Infinispan over Ehcache). Another one is startup time (our server is ready to go in 5-6 seconds opposed to 1 minute in JBoss 5). Finally – development is much faster (EJB 3.1 is much better then 3.0). The last one might be translated to “time to market”. Having above arguments I'm pretty sure you'll convince them. Step #2 – Do some reading Here is a list on interesting links which are worth reading before the migration: JBoss 5 -> 7 migration guide: https://docs.jboss.org/author/display/AS7/How+do+I+migrate+my+application+from+AS5+or+AS6+to+AS7 JBoss 7 vs EAP libraries: https://access.redhat.com/site/articles/112673 JBoss EAP Faq: http://www.jboss.org/jbossas/faq Cache implementation benchmarks: http://sourceforge.net/p/nitrocache/blog/2012/05/performance-benchmark-nitrocache--ehcache--infinispan--jcs--cach4j/ JBoss 7 performence tuning: http://www.mastertheboss.com/jboss-performance/jboss-as-7-performance-tuning JBoss caching: http://www.mastertheboss.com/hibernate-howto/using-hibernate-second-level-cache-with-jboss-as-5-6-7 Step #3 – Off you go – change Maven dependencies JBoss 5 isn't packaged very well, so I suppose you many dependencies included in your classpath (either directly or by transitive dependencies). This is the first big change in JBoss 7. Now I strongly advice you to use this artifact in your dependency management section: org.jboss.as jboss-as-parent 7.2.0.Final pom import We also decided to stick only to JEE6 spec and configure all additional JBoss 7 options with proper XML files. If it sounds good for your project too, just add this dependency and you're done with this step: org.jboss.spec jboss-javaee-6.0 1.0.0.Final pom provided After cleaning up dependencies your code probably won't compile for a couple of days or even weeks. It takes time to clean this up. Step #4 – EJB 3.0 to 3.1 migration Dependency Injection is a heart of the application, so it is worth to start with it. Almost all of your code should work, but you'll have some problems with beans annotated with @Service (these are singletons with JBoss 5 EJB Extended API). You just need to replace them with @Singleton annotations and put @PostConstruct annotation on your init method. One last thing – remember to use proper concurrency strategy. We decided to use @ConcurrencyManagement(BEAN) and leave the implementation as is. Step #5 – Upgrade to JPA 2.0 If you used JPA 1.0 with Hibernate, I'm pretty sure you have a lot of non standard annotations defining caching or cascading. All of them might be successfully replaced with JPA 2.0 annotations and finally you might get rid of Hibernate from compile classpath and depend only on JPA 2.0. Here are several standard things to do: Get rid of Hibernate's Session.evict and switch to EntityManager.detach Get rid of Hibernate's @Cache annotation and replace it with @Cachable Fix Cascades (now delete orphan is a part of @XXXToYYY annotations) Remove Hibernate dependency and stick with JEE6 spec Step #6 – Fix Hibernate's sequencer Migrating Hibernate 3 to 4 is a bit tricky because of the way it uses sequences (fields annotated with @Id). Hibernate by default uses a pool of ids instead of incrementing sequence. An example will be more descriptive: Some_DB_Sequence.nextval -> 1 Hibernate 3: 1*50 = 50; IDs to be used = 50, 51, 52.…, 99 Some_DB_Sequence.nextval -> 2 Hibernate 3: 2*50 = 100; IDs to be used = 100, 101, 102.…, 149 In Hibernate 4.x there is a new sequence generator that uses new IDs that are 1:1 related to DB sequence. Typically it's disabled by default... but not in JBoss 7.1. So after migration, Hibernate tries to insert entities using IDs read from sequence (using new sequence generator) that were already used which causes constraint violation. The fastest solution is to switch Hibernate to the old method of sequence generation (described in example above), that requires following change in persistence.xml: Step #7 – Caching Infinispan is shipped with JBoss 7 and does not require much configuration. There is only one setting in persistence.xml which needs to be set and the others might be removed: Infinispan itself might require some extra configuration – just use standalone-full-ha.xml as guide. Step #8 – RMI with JBoss 5 If you're using a lot of RMI communicating with other JBoss 5 servers – I have bad information for you – JBoss 5 and 7 are totally different and this kind of comminication will not work. I strongly recommend to switch to some other technology like JAX-WS. In the retrospect we are very glad we decided to do it. Step #9 – JMS migration We thought it would be really hard to connect with JMS server based on JBoss 5. It turned out that you have 2 options and both work fine: Start HornetQ server on your own instance and create a bridge to JBoss 5 instance Use Generic JMS adapter: https://github.com/jms-ra/generic-jms-ra Step #10 – Fix EAR layout In JBoss 5 it does not matter where all jars are being placed. All EJBs are being started. It does not work with JBoss 7 anymore. All EJB which should start must be added as modules. Step #11 – JMX console Bad information – it's not present in JBoss 7. We liked it very much, but we had to switch to jvisualvm to invoke our JMX operations. There is a ticket in WildFly Jira opened for that: https://issues.jboss.org/browse/WFLY-1197. Unfortunately at moment of writing this article it is not resolved. Some thoughts in retrospect It is really time consuming task to migrate from JBoss 5 to 7. Although in my opinion it is worth it. Now we have better caching for cluster solutions (Infinispan), better DI (EJB 3.1) and better Web Services (CXF instead of JBoss WS). Processing time decreased by 25% without any code change. Development speed increased in my opinion (it is really hard to measure it) by 50% and we are much more productive (faster server restarts). Memory footprint lowered from 1GB to 512MB. Finally automatic application redeployment finally works! However there is always a price to pay – the migration took us 4 weeks (2 sprints). We didn't write any code for our business in that period. So make sure you prepare well for such migration and my last advice – invest some time to write good automatic functional tests (we use Arquillian for that). Once they're green again – you're almost crossing finishing line.

qunit , used by projects like jquery and jquery mobile , is a rather popular javascript testing framework. for tests written using qunit, how do we measure its code coverage ? a possible solution which is quite easy to setup is to leverage the deadly combination of karma and istanbul . just like our previous adventure with jasmine code coverage , let's take a look at some simple code we need to test. this function my.sqrt is a reimplementation of math.sqrt which may throw an exception if the input is invalid. var my = { sqrt: function(x) { if (x < 0) throw new error("sqrt can't work on negative number"); return math.exp(math.log(x)/2); } }; a very simple qunit-based test for the above code is as follows. test("sqrt", function() { deepequal(my.sqrt(4), 2, "square root of 4 is 2"); }); manually running the test is easy as opening the test runner in a web browser: for a smoothed development workflow, an automated way to run the tests will be much preferred. this is where karma becomes very useful. karma also has the ability to launch a predetermined collection of browsers, or even to use phantomjs for a pure headless execution (suitable for smoke testing and/or continuous delivery). before we can use karma, installation is necessary: npm install karma karma-qunit karma-coverage karma requires a configuration file. for this purpose, the config file is very simple. as an illustration, the execution is done by phantomjs but it is easy to include other browsers as well. module.exports = function(config) { config.set({ basepath: '', frameworks: ['qunit'], files: [ '*.js', 'test/spec/*.js' ], browsers: ['phantomjs'], singlerun: true, reporters: ['progress', 'coverage'], preprocessors: { '*.js': ['coverage'] } }); }; now you can start karma with the above configuration, it would say that the test passes just fine. should you encounter some problems, you can look at an example repository i have setup github.com/ariya/coverage-qunit-istanbul-karma , it may be useful as a starting point or a reference for your own project. as a convenience, the test in that repository can be executed via npm test . what is more interesting here is that karma runs its coverage processor, as indicated by preprocessors in the above configuration. karma will run istanbul , a full-featured instrumenter and coverage tracker. essentially, istanbul grabs the original javascript source and injects extra instrumentation code so that it can gather the execution metrics once the process finishes (read also my previous blog post on javascript code coverage with istanbul ). in this karma and istanbul combo, the generated coverage report is available in the under the subdirectory coverage . the above report indicates that the single test for my.sqrt is still missing the test for an invalid input, thanks to branch coverage feature of istanbul. the i indicator next to the conditional statement tells us that the if branch was never taken. of course, once the issue is known, adding another test which will cover that branch is easy (left as an exercise for the reader). now that code coverage is tracker, perhaps you are ready for the next level? it is about setting the hard threshold so that future coverage regression will never happen. protect yourself and your team from carelessness, overconfidence, or honest mistakes!

for modern web application development, having dozens of unit tests is not enough anymore. the actual code coverage of those tests would reveal if the application is thoroughly stressed or not. for tests written using the famous jasmine test library, an easy way to have the coverage report is via istanbul and karma . for this example, let’s assume that we have a simple library sqrt.js which contains an alternative implementation of math.sqrt . note also how it will throw an exception instead of returning nan for an invalid input. var my = { sqrt: function(x) { if (x < 0) throw new error("sqrt can't work on negative number"); return math.exp(math.log(x)/2); } }; using jasmine placed under test/lib/jasmine-1.3.1 , we can craft a test runner that includes the following spec: describe("sqrt", function() { it("should compute the square root of 4 as 2", function() { expect(my.sqrt(4)).toequal(2); }); }); opening the spec runner in a web browser will give the expected outcome: so far so good. now let's see how the code coverage of our test setup can be measured. the first order of business is to install karma . if you are not familiar with karma, it is basically a test runner which can launch and connect to a specific set of web browsers, run your tests, and then gather the report. using node.js, what we need to do is: npm install karma karma-coverage before launching karma, we need to specify its configuration . it could be as simple as the following my.conf.js (most entries are self-explained). note that the tests are executed using phantomjs for simplicity, it is however quite trivial to add other web browsers such as chrome and firefox. module.exports = function(config) { config.set({ basepath: '', frameworks: ['jasmine'], files: [ '*.js', 'test/spec/*.js' ], browsers: ['phantomjs'], singlerun: true, reporters: ['progress', 'coverage'], preprocessors: { '*.js': ['coverage'] } }); }; running the tests, as well as performing code coverage at the same time, can be triggered via: node_modules/.bin/karma start my.conf.js which will dump the output like: info [karma]: karma v0.10.2 server started at http://localhost:9876/ info [launcher]: starting browser phantomjs info [phantomjs 1.9.2 (linux)]: connected on socket n9ndnhj0np92ntspgx-x phantomjs 1.9.2 (linux): executed 1 of 1 success (0.029 secs / 0.003 secs) as expected (from the previous manual invocation of the spec runner), the test passed just fine. however, the most particular interesting piece here is the code coverage report, it is stored (in the default location) under the subdirectory coverage . open the report in your favorite browser and there you'll find the coverage analysis report. behind the scene, karma is using istanbul , a comprehensive javascript code coverage tool (read also my previous blog post on javascript code coverage with istanbul ). istanbul parses the source file, in this example sqrt.js , using esprima and then adds some extra instrumentation which will be used to gather the execution statistics. the above report that you see is one of the possible outputs, istanbul can also generate lcov report which is suitable for many continuous integration systems (jenkins, teamcity, etc). an extensive analysis of the coverage data should also prevent any future coverage regression, check out my other post hard thresholds on javascript code coverage . one important thing about code coverage is branch coverage . if you pay attention carefully, our test above is still not exercising the situation where the input to my.sqrt is negative. there is a big "i" marking in the third-line of the code, this is istanbul telling us that the if branch is not taken at all (for the else branch, it will be an "e" marker). once this missing branch is noticed, improving the situation is as easy as adding one more test to the spec: it("should throw an exception if given a negative number", function() { expect(function(){ my.sqrt(-1); }). tothrow(new error("sqrt can't work on negative number")); }); once the test is executed again, the code coverage report looks way better and everyone is happy. if you have some difficulties following the above step-by-step instructions, take a look at a git repository i have prepared: github.com/ariya/coverage-jasmine-istanbul-karma . feel free to play with it and customize it to suit your workflow!

This post comes from Michael Rice at the NuoDB Techblog. Tuning Linux I/O Scheduler for SSDs Hello Techblog readers! I'm going to talk about tuning the Linux I/O scheduler to increase throughput and decrease latency on an SSD. I'll also cover another interesting topic about tuning the performance of NuoDB by specifying storage for the archive and journal directories. Tuning the Linux I/O Scheduler Linux gives you the option to select the I/O scheduler. The scheduler can be changed without rebooting, too! You may be asking at this point, "why would I ever want to change the I/O scheduler?" Changing the scheduler makes sense when the overhead of optimizing the I/O (re-ordering I/O requests) is unnecessary and expensive. This setting should be fine-tuned per storage device. The best setting for an SSD will not be a good setting for an HDD. The current I/O scheduler can be viewed by typing the following command: mrice@host:~$ cat /sys/block/sda/queue/scheduler noop anticipatory deadline [cfq] The current I/O scheduler (in brackets) for /dev/sda on this machine is CFQ, Completely Fair Queuing. This setting became the default in kernel 2.6.18 and it works well for HDDs. However, SSDs don't have rotating platters or magnetic heads. The I/O optimization algorithms in CFQ don't apply to SSDs. For an SSD, the NOOP I/O scheduler can reduce I/O latency and increase throughput as well as eliminate the CPU time spent re-ordering I/O requests. This scheduler typically works well on SANs, SSDs, Virtual Machines, and even fancy Fusion I/O cards. At this point you're probably thinking "OK, I'm sold! How do I change the scheduler already?" You can use the echo command as shown below: mrice@host:~$ echo noop | sudo tee /sys/block/sda/queue/scheduler To see the change, just cat the scheduler again. mrice@host:~$ cat /sys/block/sda/queue/scheduler [noop] anticipatory deadline cfq Notice that noop is now selected in brackets. This change is only temporary and will reset back to the default scheduler, CFQ in this case, when the machine reboots. You need to edit the Grub configuration in order to keep the setting permanently. However, this will change the I/O scheduler for all block devices. The problem is that NOOP is not a good choice for HDD. I'd only permanently change the setting if the machine only has SSDs. On Grub 2: Edit: /etc/default/grub Add "elevator=noop" to the GRUB_CMDLINE_LINUX_DEFAULT line. sudo update-grub At this point, you've changed the I/O scheduler to NOOP. How do you know if it made a difference? You could run a benchmark against it and compare the numbers (just remember to flush the file-system cache). The other way is to take a look at the output from iostat. I/O requests spend less time in the queue with the NOOP I/O scheduler. This can be seen with in the "await" field from iostat. Here's an example of a larger write operation with NOOP. iostat -x 1 /dev/sda Device: sda rrqm/s 0.00 wrqm/s 143819.00 r/s 6.00 w/s 2881.00 rkB/s 24.00 wkB/s 586800.00 avgrq-sz 406.53 avgqu-sz 0.94 await 0.33 r_await 3.33 w_await 0.32 svctm 0.11 %util 31.00 Tuning NuoDB Performance Now that you've learned about the NOOP I/O scheduler, I'll talk about tuning NuoDB with an SSD. If you've read the tech blogs you'll know that there are two building blocks for a NuoDB database: the Transaction Engine, TE for short, and the Storage Manager, SM. The TE is an in-memory only copy of the database (actually a portion of the database). As a result, an SSD won't help the performance of a TE because, it doesn't store atoms to disk. The SM contains two modules that write to disk: the archive and the journal. The archive stores atoms to disk when archive configuration parameter points to a file system (versus HDFS and S3). The journal, on the other hand, synchronously writes messages to disk. If you read the blog post on durability, you may remember that the "Remote Commit with Journaling" setting provides the highest level of durability but at the cost of slower speed. Using an SSD in this situation can drastically improve performance. To tune this setting, we'll need to make a nuodb directory on the SSD: mkdir -p /ssd/nuodb The SSD in this example has the mount point /ssd on this machine. Easy, right? I'm assuming you've already set the Linux I/O scheduler for the SSD to NOOP. The next step is to configure NuoDB to use this path when creating the journal directory. The journal has a direct correlation on the transaction throughput because the journal has to finish the disk write before an ACK for the transaction commit is sent by the SM back to the TE. What about the archive? The archive is decoupled from the transaction commit, the atoms will remain in memory and gradually make their way to disk. This will have very little effect on the TPS of the database. As a result, the archive directory can be placed on just a regular HDD. The quick guide for tuning performance is to put the journal on an SSD and archive can be placed on an HDD. Here are the commands: nuodbmgr --broker localhost --password bird nuodb [domain] > start process sm Database: hockey Host: s1 Process command-line options: --journal enable --journal-dir /ssd/nuodb/demo-journal Archive directory: /var/opt/nuodb/demo-archives Initialize archive: true Started: [SM] s1/127.0.0.1:37880 [ pid = 25036 ] ACTIVE nuodb [domain/hockey] > start process te Host: t1 Process command-line options: --dba-user dba --dba-password goalie Started: [TE] t1/127.0.0.1:48315 [ pid = 25052 ] ACTIVE The important point with this tuned configuration is that only the journal is on an SSD. As a result, the SSD doesn't have to be one of these massive TB drives. The costs of SSDs have significantly dropped in price and a single 128 GB or 256 GB SSD would be adequate for the journal data. This configuration should rival the local commit performance which is wicked awesome considering it is the highest durability level! I encourage you to try it out and ask some questions.

Sharding is a database partitioning technique that distributed Aggregates such as rows or documents across multiple servers; this choice for horizontal queries trades in some client complexity (whose queries must include a shard key such as a zip code or a customer id) for the capability of distributing the dataset between multiple servers, scaling not only the read but also the write capacity. On the application side, there are several singleton processes - for example cron configurations - that are usually run only once on the whole data set. For a certain category of singleton processes we can switch to a shard-like architecture that can scale first to multiple processes and when necessary to multiple servers. Step 1: identify the candidate process Take a look at your crontab or at your process scheduling configuration if you use another infrastructure. Some of the processes are aggregations of data producing statistics, and their work can already be distributed with patterns such as MapReduce. The kind of processes interesting for client sharding is the one where an operation is performed over every single element of the data set. Each element is an Aggregate and as such does not interact, in a single transaction, with other ones. Therefore these processes are intrinsically parallelizable: you only need a way to distribute the load. Some examples of shard-able processes are: rebuild the data aggregates for a new day for each user perform some consistency checks on every customer order send all pending orders perform a renewal for each user subscription Step 2: choose a shard key Once you have identified the aggregates along which to parallelize a length operation, the choice of the shard key will usually be straightforward. This key must be uniformly distributed between the N shards you want to create on the client side. Some examples: the zip code for customers the numerical, sequential id for orders a UUID for uploaded videos the transaction reference number for money transactions Step 3: divide the work with the shard key A process before the application of sharding is usually composed of two phases: select all Aggregates whose satisfy condition C apply operation O to all the selected Aggregates. The first operation can be sometime sharded directly, transforming each of the N processes in: select all Aggregates whose satisfy condition C and whose shard key is equal to this shard's number modulo N. apply operation O to all the selected Aggregates. For example, the first operation for shard 0 of 4 can be accomplished by an SQL query: SELECT * FROM aggregate_table WHERE outdated=true // condition C AND aggregate_id % 4=0// sharding Precalculating aggregate_id % 4 can improve the performance of the query, depending on your database; however it can make more difficult to rescale the number of processes. When you switch to 8 or 16 client shards it will be necessary to stop all current running processes, recalculate the column and restart the new batch. Furthermore, the performance of ALTER TABLE is usually not good on large tables which are the subject of this client sharding technique. Some times we're not able to divide the query in a partition of the original data set directly in the database. The pattern becomes: select id (and shard key if different) for all Aggregates whose satisfy condition C. filter the subset by only considering the id whose modulo N is equal to this shard's number. apply operation O to all the selected Aggregates. For example, I use this second form while using multiple client with MongoDB. I do not know if it's possible to query ObjectIds by their modulo, so I resort to selecting all of them and then filtering them out on the client side: $id = (string) $document['_id']; $numericalValue = hexdec(substr($id, -4)); // without substr() the conversion will overflow 32-bit integers if ($numericalValues % $shards == $shard) { ... } This is only useful under the assumption that is not the query that's taking too much time in the original process, but the application of the O operation to all of its results. Note also that these solutions needs well-behaving processes that do not intervene on the data of each other: the filtering is left to the programmer. In general, it is also necessary to guarantee mutual exclusion with the original singleton process; this usually comes up when deploying the battery of N crons, as they should not start until the last of the singleton processes has terminated not to be started again.

This post comes from Karan Malhi at the Mulesolft blog. Nginx (pronounced "engine X") is an HTTP and reverse proxy server. It is well known for its high performance and stability. It is pretty feature-rich and very simple to configure. Nginx hosts nearly 12.18 percent (22.2M) of active sites across all domains. Nginx uses event-driven architecture to handle requests. When compared to a thread-per-request model, event-driven is highly scalable with a low and predicatble memory footprint. Nginx is very easy to set up as a load balancer for an Apache Tomcat farm. In this blog post, I will show you how to set it up as a round-robin load balancer for two ApacheTomcat servers. You first need to install Nginx, you can find the installation instructions here. If you are on a Mac, then you could use Homebrew brew install nginx After installing, you can run it using the nginx command sudo nginx You can now test the installation by opening the following URL in a browser: http://localhost:8080 Lets change the default port 8080 to port 80. You can do that in the nginx.conf file which by default is located at /usr/local/etc/nginx/nginx.conf. First, stop nginx: sudo nginx -s stop Now, open the nginx.conf file and locate the listen attribute of the server. Change the value from 8080 to 80. Here is where I made the change in my configuration: server { listen 80; server_name localhost; Save the file and start nginx. You should now be able to test the configuration change by visiting http://localhost Install two instances of Apache Tomcat. Open their server.xml files and change the HTTP port numbers to 8080 and 8081 respectively. You can find more information on Apache Tomcat configuration here. Once you have made the changes and saved the configuration files, you should now start each instance of Apache Tomcat. Here are a few ways you can start Apache Tomcat. Typically you should be able to locate the bin directory inside your Tomcat installation and invoke the startup.sh file as shown: bin/startup.sh Now that the Tomcat instances are started, let's set up the round robin load balancer. You would need to use the Nginx upstream module. The upstream module allows you to group servers that can be referenced by the proxy_pass directive. In your nginx.conf, locate the http block and add the upstream block to create a group of servers: http { upstream tomcat_servers{ ip_hash; server 127.0.0.1:8080; server 127.0.0.1:8081; } ip_hash above configures the load balancing method where requests are routed to servers based on IP Addresses of the client, so a request from a client with a particular IP will always go to the same back end Tomcat Server. Finally, you need to locate the location block within the server block and map the root(/) location to the tomcat_servers group created using the upstream module above. location / { proxy_pass http://tomcat_servers; } That's it!. Restart Nginx and you should now be able to send a request to http://localhost and the request will be served by one of the Tomcat servers.

One of the five essential attributes of cloud computing (see The 5-3-2 Principle of Cloud Computing) is resource pooling, which is an important differentiator separating the thought process of traditional IT from that of a service-based, cloud computing approach. Resource pooling in the context of cloud computing and from a service provider’s viewpoint denotes a set of strategies and a methodical way of managing resources. For a user, resource pooling institutes an abstraction for presenting and consuming resources in a consistent and transparent fashion. This article presents these key concepts derived from resource pooling: Resource Pools Virtualization in the Context of Cloud Computing Standardization, Automation, and Optimization Fabric Cloud Closing Thoughts Resource Pools Ultimately, data center resources can be logically placed into three categories. They are: compute, networks, and storage. For many, this grouping may appear trivial. It is, however, a foundation upon which some cloud computing methodologies are developed, products designed, and solutions formulated. Compute This is a collection of all CPU capabilities. Essentially all data center servers, either for supporting or actually running a workload, are all part of this compute group. Compute pool represents the total capacity for executing code and running instances. The process to construct a compute pool is to first inventory all servers and identify virtualization candidates followed by implementing server virtualization. It is never too early to introduce a system management solution to facilitate the processes, which in my view is a strategic investment and a critical component for all cloud initiatives. Networks The physical and logical artifacts putting in place to connect resources, segment, and isolate resources from layer three and below, etc., are gathered in the network pool. Networking enables resources becoming visible and hence possibly manageable. In the age of instant gratification, networks and mobility are redefining the security and system administration boundaries, and play a direct and impactful role in user productivity and customer satisfaction. Networking in cloud computing is more than just remote access, but empowerment for a user to self-serve and consume resources anytime, anywhere, with any device. BYOD and consumerization of IT are various expressions of these concepts. Storage This has long been a very specialized and sometimes mysterious part of IT. An enterprise storage solution is frequently characterized as a high-cost item with a significant financial and contractual commitment, specialized hardware, proprietary API and software, a dependency on direct vendor support, etc. In cloud computing, storage has become even more noticeable since the ability to grow and shrink based on demands, i.e. elasticity, demands an enterprise-level, massive, reliable, and resilient storage solution at a global scale. While enterprise IT is consolidating resources and transforming the existing establishment into a cloud computing environment, how to leverage existing storage devices from various vendors and integrate them with the next generation storage solutions is among the highest priorities for modernizing a data center. Virtualization in the Context of Cloud Computing In the last decade, virtualization has proved its value and accelerated the realization of cloud computing. Then, virtualization was mainly server virtualization, which in an over-simplified statement means hosting multiple server instances with the same hardware while each instance runs transparently and in insolation, as if each consumes the entire hardware and is the only instance running. Much of the customer expectations, business needs, and methodologies has since evolved. Now, we should validate virtualization in the context of cloud computing to fully address the innovations rapidly changing how IT conducts business and delivers services. As discussed below, in the context of cloud computing, consumable resources are delivered in some virtualized form. Various virtualization layers collectively construct and form the so-called fabric. Server Virtualization The concept of server virtualization remains: running multiple server instances with the same hardware while each instance runs transparently and in isolation, as if each instance is the only instance running and consuming the entire server hardware. In addition to virtualizing and consolidating servers, server virtualization also signifies the practices of standardizing server deployment switching away from physical boxes to VMs. Server virtualization is for packaging, delivering, and consuming a compute pool. There are a few important considerations of virtualizing servers. IT needs the ability to identify and manage bare metal such that the entire resource life-cycle management from commencing to decommissioning can be standardized and automated. To fundamentally reduce the support and training cost while increasing productivity, a consistent platform with tools applicable across physical, virtual, on-premises, and off-premises deployments is essential. The last thing IT wants is one set of tools for physical resources and another for those virtualized, one set of tools for on-premises deployment and another for those deployed to a service provider, and one set of tools for development and another for deploying applications. The requirement is one methodology for all, one skill set for all, and one set of tools for all. This advantage is obvious when developing applications and deploying Windows Server 2012 R2 on premises or off premises to Windows Azure. The Active Directory security model can work across sites, System Center can manage resources deployed off premises to Windows Azure, and Visual Studio can publish applications across platforms. Windows infrastructure architecture, security, and deployment models are all directly applicable. Network Virtualization The similar idea of server virtualization applies here. Network virtualization is the ability to run multiple networks on the same network device while each network runs transparently and in isolation, as if each network is the only network running and consuming the entire network hardware. Conceptually, since each network instance is running in isolation, one tenant’s 192.168.x network is not aware of another tenant’s identical192.168.x network running with the same network device. Network virtualization provides the translation between physical network characteristics and the representation of and a resource identity in a virtualized network. Consequently, above the network virtualization layer, various tenants while running in isolation can have identical network configurations. A great example of network virtualization is Windows Azure virtual networking. At any given time, there can be multiple Windows Azure subscribers all allocating the same 192.168.x address space with an identical subnet scheme (192.168.1.x/16) for deploying VMs. Those VMs belonging to one subscriber will however not be aware of or visible to those deployed by others, despite the fact that the network configuration, IP scheme, and IP address assignments may all be identical. Network virtualization in Windows Azure isolates on subscriber from the others such that each subscriber operates as if the subscription is the only one employing a 192.168.x address space. Storage Virtualization I believe this is where the next wave of drastic cost reduction of IT post-server virtualization happens. Historically, storage has been a high cost item in any IT budget in each and every aspects including hardware, software, staffing, maintenance, SLA, etc. Since the introduction of Windows Server 2012, there is a clear direction where storage virtualization is built into OS and becoming a commodity. New capabilities like Storage Pool, Hyper-V over SMB, Scale-Out Fire Share, etc., are now part of Windows Server OS and are making storage virtualization part of server administration routines and easily manageable with tools and utilities like PowerShell, which is familiar to many IT professionals. The concept of storage virtualization remains consistent with the idea of logically separating a computing object from its hardware, in this case the storage capacity. Storage virtualization is the ability to integrate multiple and heterogeneous storage devices, aggregate the storage capacities, and present/manage as one logical storage device with a continuous storage space. JBOD is a technology to realize this concept. Standardization, Automation and Optimization Each of the three resource pools has an abstraction to logically present itself with characteristics and work patterns. A compute pool is a collection of physical (virtualization and infrastructure) hosts and VMs. A virtualization host hosts VMs that run workloads deployed by service owners and consumed by authorized users. A network pool encompasses network resources including physical devices, logical switches, address spaces, and site configurations. Network virtualization as enabled/defined in configurations can identify and translate a logical/virtual IP address into a physical one, such that tenants with the same network hardware can implement an identical network scheme without a concern. A storage pool is based on storage virtualization which is a concept of presenting an aggregated storage capacity as one continuous storage space as if provided from one logical storage device. In other words, the three resource pools are wrapped with server virtualization, network virtualization, and storage virtualization, respectively. Each virtualization presents a set of methodologies on which work patterns are derived and common practices are developed. These virtualization layers provides opportunities to standardize, automate, and optimize deployments and considerably facilitates the adoption of cloud computing. Standardization Virtualizing resources decouples the dependency between instances and the underlying hardware. This offers an opportunity to simplify and standardize the logical representation of a resource. For instance, a VM is defined and deployed with a VM template that provides a level of consistency with a standardized configuration. Automation Once VM characteristics are identified and standardized, we can now generate an instance by providing only instance-based information or information that depends on run-time, such as the VM machine name, which must be validated at run-time to prevent duplicated names. This requirement for providing only minimal information at deployment can be significantly simplify and streamline operations for automation. And with automation, resources can then be deployed, instantiated, relocated, taken off-line, brought back online, or removed rapidly and automatically based on set criteria. Standardization and automation are essential mechanisms so that workload can be scaled on demand, i.e., become elastic. Optimization Standardization provides a set of common criteria. Automation executes operations based on set criteria with volumes, consistency, and expediency. With standardization and automation, instances can be instantiated with consistency, efficiency, and predictability. In other words, resources can be operated in bulk with consistency and predictability. The next logical step is then to optimize the usage based on SLA. The presented progression is what resource pooling and virtualizations can provide and facilitate. These methodologies are now built into products and solutions. Windows Server 2012 R2 and System Center 2012 and later integrate server virtualization, network virtualization, and storage virtualization into one consistent solution platform with standardization, automation, and optimization for building and managing clouds. Fabric This is a significant abstraction in cloud computing. Fabric implies accessibility and discoverability, and denotes the ability to discover, identify, and manage a resource. Conceptually, fabric is an umbrella term encompassing all the underlying infrastructure supporting a cloud computing environment. At the same time, a fabric controller represents the system management solution which manages, i.e. owns, fabric. In cloud architecture, fabric consists of the three resource pools: compute, networks, and storage. Compute provides the computing capabilities, executes code, and runs instances. Networks glues the resources based on requirements. Storage is where VMs, configurations, data, and resources are kept. Fabric shields the physical complexities of the three resource pools presented with server virtualization, network virtualization, and storage virtualization. All operations are eventually directed by the fabric controller of a data center. Above fabric, there are logical views of consumable resources including VMs, virtual networks, and logical storage drives. By deploying VMs, configuring virtual networks, or acquiring storage, a user consumes resources. Under fabric, there are virtualization and infrastructure hosts, Active Directory, DNS, clusters, load balancers, address pools, network sites, library shares, storage arrays, topology, racks, cables, etc., all under the fabric controller’s command to collectively present and support fabric. For a service provider, building a cloud computing environment is essentially establishing a fabric controller and constructing fabric. Namely, instituting a comprehensive management solution, building the three resource pools, and integrating server virtualization, network virtualization, and storage virtualization to form fabric. From a user’s point of view, how and where a resource is physically provided is not a concern, but the accessibility, readiness, scalability, and fulfillment of SLA are. Cloud This is a well-defined term and we should not be confused with it. (see NIST SP 800-145 and the 5-3-2 Principle of Cloud Computing) We need to be very clear on: what a cloud must exhibit (the five essential attributes), how to consume it (with SaaS, PaaS, or IaaS), and the model a service is deployed in (like private cloud, public cloud, and hybrid cloud). Cloud is a concept, a state, a set of capabilities such that a business can be delivered as a service, i.e. available on demand. The architecture of a cloud computing environment is presented with three resource pools: compute, networks, and storage. Each is an abstraction provided by a virtualization layer. Server virtualization presents a compute pool with VMs that supply the computing, i.e. CPUs, and power to execute code and run instances. Network virtualization offers a network pool and is the mechanism that allows multiple tenants with identical network configurations on the same virtualization host while connecting, segmenting, isolating network traffic with virtual NICs, logical switches, address space, network sites, IP pools, etc. Storage virtualization provides a logical storage device with the capacity to appear continuous and aggregated with a pool of storage devices behind the scene. The three resource pools together constitute the fabric (of a cloud) while the three virtualization layers collectively form the abstraction, such that while the underlying physical infrastructure may be intricate, the user experience above fabric remains logical and consistent. Deploying a VM, configuring a virtual network, or acquiring storage is transparent with virtualization regardless of where the VM actually resides, how the virtual network is physically wired, or what devices in the aggregate the requested storage is provided with. Closing Thoughts Cloud is a very consumer-focused approach. It is about a customer’s ability and control based on SLA in getting resources when needed and with scale, and equally important releasing resources when no longer required. It is not about products and technologies. It is about servicing, consuming, and strengthening the bottom line.

A recent Java production performance problem forced me to revisit and truly appreciate the Java VM Just-In-Time (JIT) compiler. Most Java developers and support individuals have heard of this JVM run time performance optimization but how many truly understand and appreciate its benefits? This article will share with you a troubleshooting exercise I was involved with following the addition of a new virtual server (capacity improvement and horizontal scaling project). For a more in-depth coverage of JIT, I recommend the following articles: ## Just-in-time compilation http://en.wikipedia.org/wiki/Just-in-time_compilation ## The Java HotSpot Performance Engine Architecture http://www.oracle.com/technetwork/java/whitepaper-135217.html ## Understanding Just-In-Time Compilation and Optimization http://docs.oracle.com/cd/E15289_01/doc.40/e15058/underst_jit.htm ## How the JIT compiler optimizes code http://pic.dhe.ibm.com/infocenter/java7sdk/v7r0/index.jsp?topic=%2Fcom.ibm.java.zos.70.doc%2Fdiag%2Funderstanding%2Fjit_overview.html JIT compilation overview The JIT compilation is essentially a process that improves the performance of your Java applications at run time. The diagram below illustrates the different JVM layers and interaction. It describes the following high level process: Java source files are compiled by the Java compiler into platform independent bytecode or Java class files. After your fire your Java application, the JVM loads the compiled classes at run time and execute the proper computation semantic via the Java interpreter. When JIT is enabled, the JVM will analyze the Java application method calls and compile the bytecode (after some internal thresholds are reached) into native, more efficient, machine code. The JIT process is normally prioritized by the busiest method calls first. Once such method call is compiled into machine code, the JVM executes it directlyinstead of “interpreting” it. The above process leads to improved run time performance over time. Case study Now here is the background on the project I was referring to earlier. The primary goal was to add a new IBM P7 AIX virtual server (LPAR) to the production environment in order to improve the platform’s capacity. Find below the specifications of the platform itself: Java EE server: IBM WAS 6.1.0.37 & IBM WCC 7.0.1 OS: AIX 6.1 JDK: IBM J2RE 1.5.0 (SR12 FP3 +IZ94331) @64-bit RDBMS: Oracle 10g Platform type: Middle tier and batch processing In order to achieve the existing application performance levels, the exact same hardware specifications were purchased. The AIX OS version and other IBM software’s were also installed using the same version as per existing production. The following items (check list) were all verified in order to guarantee the same performance level of the application: Hardware specifications (# CPU cores, physical RAM, SAN…). OS version and patch level; including AIX kernel parameters. IBM WAS & IBM WCC version, patch level; including tuning parameters. IBM JRE version, patch level and tuning parameters (start-up arguments, Java heap size…). The network connectivity and performance were also assessed properly. After the new production server build was completed, functional testing was performed which did also confirm a proper behaviour of the online and batch applications. However, a major performance problem was detected on the very first day of its production operation. You will find below a summary matrix of the performance problems observed. Production server Operation elapsed time Volume processed (# orders) CPU % (average) Middleware health Existing server 10 hours 250 000 (baseline) 20% healthy *New* server 10 hours 50 000 -500% 80% +400% High thread utilization As you can see from the above view, the performance results were quite disastrous on the first production day. Not only much less orders were processed by the new production server but the physical resource utilization such as CPU % was much higher compared with the existing production servers. The situation was quite puzzling given the amount of time spent ensuring that the new server was built exactly like the existing ones. At that point, another core team was engaged in order to perform extra troubleshooting and identify the source of the performance problem. Troubleshooting: searching for the culprit... The troubleshooting team was split in 2 in order to focus on the items below: Identify the source of CPU % from the IBM WAS container and compare the CPU footprint with the existing production server. Perform more data and file compares between the existing and new production server. In order to understand the source of the CPU %, we did perform an AIX CPU per Thread analysis from the IBM JVM running IBM WAS and IBM WCC. As you can see from the screenshot below, many threads were found using between 5-20% each. The same analysis performed on the existing production server did reveal fewer # of threads with CPU footprint always around 5%. Conclusion: the same type of business process was using 3-4 times more CPU vs. the existing production server. In order to understand the type of processing performed, JVM thread dumps were captured at the same time of the CPU per Thread data. Now the first thing that we realized after reviewing the JVM thread dump (Java core) is that JIT was indeed disabled! The problem was also confirmed by running the java –version command from the running JVM processes. This finding was quite major, especially given that JIT was enabled on the existing production servers. Around the same time, the other team responsible of comparing the servers did finally find differences between the environment variables of the AIX user used to start the application. Such compare exercise was missed from the earlier gap analysis. What they found is that the new AIX production server had the following extra entry: JAVA_COMPILER=NONE As per the IBM documentation, adding such environment variable is one of the ways to disableJIT. Complex root cause analysis, simple solution In order to understand the impact of disabling JIT in our environment, you have to understand its implication. Disabling JIT essentially means that the entire JVM is now running ininterpretation mode. For our application, running in full interpretation mode not only reduces the application throughput significantly but also increases the pressure point on the server CPU utilization since each request/thread takes 3-4 more times CPU than a request executed with JIT (remember, when JIT is enabled, the JVM will perform many calls to the machine/native code directly). As expected, the removal of this environment variable along with the restart of the affected JVM processes did resolve the problem and restore the performnance level. Assessing the JIT benefits for your application I hope you appreciated this case study and short revisit of the JVM JIT compilation process. In order to understand the impact of not using JIT for your Java application, I recommend that you preform the following experiment: Generate load to your application with JIT enabled and capture some baseline data such as CPU %, response time, # requests etc. Disable JIT. Redo the same testing and compare the results. I’m looking forward for your comments and please share any experience you may have with JIT.

There are two ways to speedup UNIONs in a MySQL database. First use UNION ALL if at all possible, and second try to push down your conditions. 1. UNION ALL is much faster than UNION How does a UNION work? Imagine you have two tables for shirts. The short_sleeve table looks like this: blue green gray black And long_sleeve another that looks like this: red green yellow blue Related: Why Generalists are Better at Scaling the Web If you UNION those two tables, first MySQL will sort the combined set into a temp table like this: black blue blue gray green green red yellow Once it’s done this sort, it can easily remove the duplicate blue & duplicate green for this resulting set: black blue gray green red yellow See also: Mythical MySQL DBA – the talent drought. Why does it do this? UNION is defined that way in SQL. Duplicates must be removed and this is an efficient way for the MySQL engine to remove them. Combine results, sort, remove duplicates and return the set. Queries with UNION can be accelerated in two ways. Switch to UNION ALL or try to push ORDER BY, LIMIT and WHERE conditions inside each subquery. You’ll be glad you did! What if we did UNION ALL? The result would look like this: blue green gray black red green yellow blue Read this: MySQL DBA Interview & Hiring Guide. It doesn’t have to sort, and doesn’t have to remove duplicates. If you imagine combining two 10 million row tables, and don’t have to sort, this speedup can be HUGE. 2. Use Push-down Conditions to speedup UNION in MySQL Imagine with our example above the shirts have a design date, the year they were released. Yes we’re keeping this example very simple to illustrate the concept. Here is the short_sleeve table: blue 2013 green 2013 green 2012 gray 2011 black 2009 black 2011 And long_sleeve table looks like this: red 2012 red 2013 green 2011 yellow 2010 blue 2011 For 2013 designs could combine them like this: (SELECT type, release FROM short_sleeve) UNION (SELECT type, release FROM long_sleeve); WHERE release >=2013; See also: 5 More Things Deadly to Scalability and the original 5 Things Toxic to Scalability.. Here the WHERE clause works on this 11 record temp table: black 2009 black 2011 blue 2011 blue 2013 gray 2011 green 2013 green 2012 green 2011 red 2012 red 2013 yellow 2010 But it would be much faster to move the WHERE inside each subquery like this: (SELECT type, release FROM short_sleeve WHERE release >=2013) UNION (SELECT type, release FROM long_sleeve WHERE release >=2013); That would be operating on a combined 3 record table. Faster to sort & remove duplicates. Smaller result sets cache better too, providing a pay forward dividend. That’s what performance optimization is all about! Read this: RDS or MySQL – 10 Use Cases. Remember multi-million row sets in each part of this query will quickly illustrate the optimization. We’re using very small results to make visualizing easier. You can also use this optimization for ORDER BY and for LIMIT conditions. By reducing the number of records returned by EACH PART of the UNION, you reduce the work that happens at the stage where they are all combined. If you’re seeing some UNION queries in your slow query log, I suggest you try this optimization out and see if you can tweak it.

A typical ASP.NET public website hosted on IIS is usually configured in such a way that the server that IIS is installed on is visible to the public internet. HTTP requests from a browser or web service client are routed directly to IIS which also hosts the ASP.NET worker process. All the functionality needed to produce the web site is embodied in a single server. This includes caching, SSL termination, authentication, serving static files and compression. This approach is simple and straightforward for small sites, but is hard to scale, both in terms of performance, and in terms of managing the complexity of a large complex application. This is especially true if you have a distributed service oriented architecture with multiple HTTP endpoints that appear and disappear frequently. A reverse proxy is server component that sits between the internet and your web servers. It accepts HTTP requests, provides various services, and forwards the requests to one or many servers. Having a point at which you can inspect, transform and route HTTP requests before they reach your web servers provides a whole host of benefits. Here are some: Load Balancing This is the reverse proxy function that people are most familiar with. Here the proxy routes incoming HTTP requests to a number of identical web servers. This can work on a simple round-robin basis, or if you have statefull web servers (it’s better not to) there are session-aware load balancers available. It’s such a common function that load balancing reverse proxies are usually just referred to as ‘load balancers’. There are specialized load balancing products available, but many general purpose reverse proxies also provide load balancing functionality. Security A reverse proxy can hide the topology and characteristics of your back-end servers by removing the need for direct internet access to them. You can place your reverse proxy in an internet facing DMZ, but hide your web servers inside a non-public subnet. Authentication You can use your reverse proxy to provide a single point of authentication for all HTTP requests. SSL Termination Here the reverse proxy handles incoming HTTPS connections, decrypting the requests and passing unencrypted requests on to the web servers. This has several benefits: Removes the need to install certificates on many back end web servers. Provides a single point of configuration and management for SSL/TLS Takes the processing load of encrypting/decrypting HTTPS traffic away from web servers. Makes testing and intercepting HTTP requests to individual web servers easier. Serving Static Content Not strictly speaking ‘reverse proxying’ as such. Some reverse proxy servers can also act as web servers serving static content. The average web page can often consist of megabytes of static content such as images, CSS files and JavaScript files. By serving these separately you can take considerable load from back end web servers, leaving them free to render dynamic content. Caching The reverse proxy can also act as a cache. You can either have a dumb cache that simply expires after a set period, or better still a cache that respects Cache-Control and Expires headers. This can considerably reduce the load on the back-end servers. Compression In order to reduce the bandwidth needed for individual requests, the reverse proxy can decompress incoming requests and compress outgoing ones. This reduces the load on the back-end servers that would otherwise have to do the compression, and makes debugging requests to, and responses from, the back-end servers easier. Centralised Logging and Auditing Because all HTTP requests are routed through the reverse proxy, it makes an excellent point for logging and auditing. URL Rewriting Sometimes the URL scheme that a legacy application presents is not ideal for discovery or search engine optimisation. A reverse proxy can rewrite URLs before passing them on to your back-end servers. For example, a legacy ASP.NET application might have a URL for a product that looks like this: http://www.myexampleshop.com/products.aspx?productid=1234 You can use a reverse proxy to present a search engine optimised URL instead: http://www.myexampleshop.com/products/1234/lunar-module Aggregating Multiple Websites Into the Same URL Space In a distributed architecture it’s desirable to have different pieces of functionality served by isolated components. A reverse proxy can route different branches of a single URL address space to different internal web servers. For example, say I’ve got three internal web servers: http://products.internal.net/ http://orders.internal.net/ http://stock-control.internal.net/ I can route these from a single external domain using my reverse proxy: http://www.example.com/products/ -> http://products.internal.net/ http://www.example.com/orders/ -> http://orders.internal.net/ http://www.example.com/stock/ -> http://stock-control.internal.net/ To an external customer it appears that they are simply navigating a single website, but internally the organisation is maintaining three entirely separate sites. This approach can work extremely well for web service APIs where the reverse proxy provides a consistent single public facade to an internal distributed component oriented architecture. So … So, a reverse proxy can off load much of the infrastructure concerns of a high-volume distributed web application. We’re currently looking at Nginx for this role. Expect some practical Nginx related posts about how to do some of this stuff in the very near future. Happy proxying!