The Death of WebSphere and WebLogic App Servers? New Infographic shows the Rise of OSS Java
This past September we released one of our Ruby on Rails State of the Stack reports, which presents stats on Ruby usage among our customers. The report shows the most commonly deployed versions of Ruby, the top gems used, the most commonly used app containers, etc. We got to thinking that this might be interesting data to provide for other languages that we support too, such as Java. So…we decided to take a look at our top 1000 Java customers to see what versions of Java they are using and what app servers are most commonly deployed. For example, just about everyone is on Java 1.6 with the most popular version being 1.6.0_20 followed closely by 1.6.0_26. What about app servers? What are folks using? The answer was intriguing so we thought we’d share it via a new infographic…
By the way, we use Jetty for our java components. :)
Sign up here. It's free, so why not?
17.98% for “other” seems a little high especially when you list Weblogic at 0.5%.
Posted: 10 January 2012 at 8:55 am by Mat
It’s sad to see Glassfih with so little market-share because it is both fast and feature-rich
Posted: 10 January 2012 at 1:42 pm by rds
This is a self-selecting result and a sensationalist blog title , but it’s your blog
IBM & Oracle have rich tooling built in to WebSphere & WebLogic, and additional tooling available beyond that (such as Oracle Enterprise Manager & Tivoli). Tomcat has no built-in tooling in this area.
For production deployments, Tomcat users have no choice but to go to 3rd parties like New Relic. Oracle & IBM customers only need to do so to fill required gaps given their respective product’s inherent rich feature set in this area. Many Oracle and IBM customers will likely tend towards Oracle & IBM value-add monitoring/mgmt products (Oracle EM & Tivoli). They do this to reduce the number of vendors in their environment and/or to get better pricing by bundling more products.
This blog could be titled, “New Relic not offering enough value differentiation for WebLogic & WebSphere customers”
Now, can we turn the table and ask what is the market share for New Relic?
Lew Cirne Reply:
January 10th, 2012 at 8:20 pm
John, thanks for your comment. First, let me say that I think Glassfish is a great app server and I bet it will continue to gain share as more applications move to next generation open source application servers.
Our blog post represents data that we see from our customer accounts. We collect 8 billion metrics a day from our customer base and the sample that we used in this study is a very real indicator of the market conditions that we are seeing, from small startups all the way to large enterprises. In the past I have written here, here, and here on why the old system is broken and why a new way of approaching the market is needed–a way that puts the customer first by providing immediate time to value at an affordable cost, while providing a rich feature set that far exceeds customers’ expectations. I have said time and again that if you provide these capabilities in this way, your customers will love your product. And that’s how our customers feel about New Relic.
Your comment strikes me as a classic example of the old-school, enterprise software mentality. For decades, the IBM and Oracle approach to doing business has been to provide a hodgepodge of expensive, outdated tools assembled through acquisitions, and present them as an “integrated solution”, and to leverage the fact that they have a chokehold on their relationship with the CIO. From my perspective, at old school enterprise software companies, innovation is rare and the business strategy appears to focus on customer lockin rather than victory through product excellence.
New Relic provides a simple, cost-effective solution that is second to none in functionality and TCO. We deliver new features on a weekly basis. We provide major innovations on a quarterly basis. And yet our pricing is a small fraction of the traditional on-premise alternatives. IT is waking up to this new reality: they are tired of writing big checks (plus 20% maintenance) for software that provides marginal value. That’s why world-class enterprises are moving new mission critical projects to Jetty and New Relic and as opposed to Oracle, WebLogic and OEM.
How do we know our approach works? We have 15,000 production customers, are monitoring 300,000 application instances and dominate in the Cloud (we have close partnerships with a who’s who of cloud computing platform providers). And by the way, we define a customer as an account that is actively using New Relic in production, a vastly different definition compared to the software that gets wrapped into large Oracle deals and it still sitting on the enterprise IT shelf.
–Lew Cirne, CEO
Posted: 10 January 2012 at 1:54 pm by John Clingan
Typical FUD. So what is the market share for New Relic? Don’t be shy, draw the picture.
Posted: 10 January 2012 at 9:46 pm by Ken
> Our blog post represents data that we see from our customer accounts
Which means it is not representative. Stop spreading FUD.
foljs Reply:
January 12th, 2012 at 10:50 am
What exactly is your problem?
You have emotionally invested in Oracle/IBM web application servers? Really?
Or is it Stockholm Syndrome, from being forced to use those at work?
It doesn’t take a genius to see that Open Source java stacks are winning the war. Have a look at the market OUTSIDE your corner of the corporate world.
Ken Reply:
January 20th, 2012 at 3:43 am
> Open Source java stacks are winning the war
What the war is that and what is the evidence of win? Corner is actually the place where OSS Java stacks reside. The success of Java is attributed to corporations using it. This market is much more bigger than anything occupied by “OSS Java stacks”. And IBM, Oracle, etc. dominate there.
Posted: 10 January 2012 at 9:52 pm by Ken
Maybe lumping in servlet-only servers with EJB servers doesn’t provide all that much insight. I would think its more meaningful to show the servlet-only containers with the EJB servers as one bubble, and then show a breakout of that bubble showing the relative shares of the EJB servers against one another in the breakout. I imagine people would naturally stop at servlet-only servers unless they needed full blow EJB services or if they deploy servlet-only apps on EJB servers for “consistency” because they are a [JBoss|Glassfish|Weblogic|WebSphere] shop.
Posted: 11 January 2012 at 6:52 am by Clark
I’m a fan of OSS Java, I’m glad to see a study like this. I try to increase value for my customers with OSS (both quality and price are better), so I like to show my customers these studies.
As believable as this report is, I don’t understand the math and precision on this survey. Do you have more precise stats?
For example: for 1000 responses, shouldn’t you just have 3 digits of precision? 54.16% is outstanding for Tomcat, but 4 digits seems a little too precise. I’d prefer to see 54.1%
Also, when surveying 1000 people, what number of responses for “Weblogic” will result in exactly 0.51% ? If I do the math, 5 Weblogic users out of 1000 would result in 0.500% exactly. And if you did actually get 1007 responses or some odd number like that, it seems that 0.5% is the proper number of digits.
Posted: 11 January 2012 at 8:28 am by Jay Meyer
Why are you not posting my response to your blog post??
Patrick Moran Reply:
January 11th, 2012 at 9:50 am
Hi John – I approved all of them; my apologies, we had “moderate” set to YES in wordpress… unintentional.
John Clingan Reply:
January 11th, 2012 at 9:51 am
No problem. Thanks.
Posted: 11 January 2012 at 9:42 am by John Clingan
An old way of thinking is that vendors shrink-wrap software (
), and I don’t think you have data on Oracle customer software utilization rates. Otherwise, I still think your data is self-selecting. I suppose we’ll just have to disagree, and that’s OK.
Posted: 11 January 2012 at 9:50 am by John Clingan
When looking at App servers, in the open source world. Most are based around the monolithic operating system design… They all pretty much do the same thing… I personally, if I can find the time would like to try and see what the benefits of a micro-kernal design might be in the middle ware world
John Clingan Reply:
January 11th, 2012 at 4:24 pm
Stephen, most appservers these days are modular, so you might want to have another look if you do find the time. GlassFish, for example, runs on an OSGi runtime. Take a look at the source at glassfish.org.
Posted: 11 January 2012 at 10:15 am by Stephen Lintott
My understanding is only that the Ruby customers are not WebSphere/Weblogic customers.
Not really that WebSphere/Weblogic disappear so quickly (even if I agree that non-OSS solution for appServer will disappear in mid term).
The most real % of theses app server market can be found here : http://zeroturnaround.com/java-ee-productivity-report-2011/
Patrick Moran Reply:
January 11th, 2012 at 12:40 pm
The link you provided is great, I hadn’t seen it thanks for passing along. Everyone should check that out: http://zeroturnaround.com/java-ee-productivity-report-2011/
This infographic had nothing to do with our Ruby customers; we took a look at 1,000 of our JAVA customers (we support Java, PHP, Ruby, .net, and Python). These customers include university of Phoenix, AT&T interactive, and other big (and small) customers.
Hope that helps.
Patrick
Posted: 11 January 2012 at 12:23 pm by Cédric Gérard
Its worth noting that newrelic has issues with Glassfish 2.x (can’t remember the exact version) versions. It certainly put me off in using newrelic with a Glassfish cluster early last year and had to use a different one. It was a shame as it is a great piece of software.
Natasha Reply:
January 15th, 2012 at 12:01 pm
Hi Imran,
Recent versions of New Relic’s Java agent should work with Glassfish clusters. I hope you’ll consider having another go at monitoring Glassfish with New Relic! Also, if you ever hit troubles with our agent, support at newrelic.com is standing by to help solve the problem.
Thanks,
Natasha
Imran Reply:
January 19th, 2012 at 6:14 am
When I checked with support at the time, it was said that you all didn’t have any plans on supporting the 2.x version of Glassfish. It was down to a conflict with a logging library IIRC. Anyway I do use newrelic now on tomcat now.
Posted: 12 January 2012 at 3:12 am by Imran
According to me, your statistic questionnaire had to be a bit different…..
Should use 2 questions:
1) “Which J2EE A/S do you use to develop your J2EE application ?”
2) “What is your production J2EE A/S?”
In the first question: (which seems to me that the 80% of your sample considered as the default) the default answer is Tomcat….. Obviously if a Tomcat start costs 10 secs and JBoss 6 or Websphere start is about 1.10 min and you have to test your application 1000 times, then the answer is very easy….
The second question: Now, consider a TomCat in a real demanding production line. Load Balancing, Cluster ????
I don’t think that this is a safe option….
I thing your statistic has a problem…..
Posted: 13 January 2012 at 6:36 am by Mike Mountrakis
LOL, written by a Java programmer???
Posted: 14 January 2012 at 8:45 am by web development newcastle
This survey methodology is analogous to doing a survey of which makes and models of cars show up at a BP Gas Station and than blogging about the fact that results show that there are no signs that electric cars exist in the real world.
As an earlier post stated, APM tools are also provided by the commercial app server vendors already, so it isn’t surprising that few new relic users would be using commercial app servers.
show your tools for what they are capable of and how they compete with other APM solutions, instead of drawing conclusions about an app server market from a flawed collection methodology.
Posted: 26 January 2012 at 6:11 pm by developer