rand($thoughts);

When I bought my first home, I used a lawyer that a friend referred me to.  The lawyer was great, thorough, responded quickly and took the time to explain things in plain English.

I used him again 2 years later for an investment and found that he was still thorough but not as quick to get back to me or willing to spend as much time going through the legal lingo.  Heck, I had to call him an average of 3x just to get a response from him, and it took nearly 2 years to get my closing documents.

What changed?  Well, news got out that he was good at his job.  He got busy, became overworked, and as a result, his customer service suffered.  Sadly, he couldn’t clone himself!  Also, his early customers expected a level of customer service that he was not able to give, even with additional administrative staff.

I thought about this when I was looking at the CIO study that Red Hat is linking to which indicates Red Hat was the leading vendor to work with.  NOTE: I’m not challenging the results of the study.  I do however find it interesting that the top 2 vendors listed (Red Hat & Apple) have less than the average # of responses who have worked with the vendor in the past 12 months (see 3rd column from right on the 3rd page).  In plain English, even though Red Hat & Apple scored highest in the study, less than 25% of the respondents had experience with these two vendors in the past 12 months. Yes, that means lots of room to grow for these vendors ;-)  But does it also hint at growing pains that may lie around the corner?  Said differently, I am questioning whether Red Hat’s position in the rankings will take a turn for the worse if they are able to attract additional new customers.  Will they fall into the same trap that my lawyer did?  Get too big for a service business to the point that customer satisfaction suffers severely?

Some of you will say, I missing the point; That the reason customers acquire RHEL is because it’s a higher quality product than its competition.  Well, if that is the case and the support experience doesn’t matter, then why pay for RHEL at all?  Why not use CentOS? <devilsAdvocate>Or for that matter, Oracle’s support for RHEL? </devilsAdvocate> To be fair, I’m not saying that RedHat’s success is only due to the level of support and the customer service associated with RHEL. Support, customer service and product quality are all important.  (As is vendor stability & reputation, I’d bet).

I guess I’m pointing out a scaling problem that Red Hat may encounter down the road….unless Oracle buys them ;-)  or Red Hat sees it coming and puts strategies in place to minimize the customer impact?

There is a symbiotic relationship between Oracle & Red Hat now. Yes there was one before with Oracle’s database business on Linux, but it’s even more intertwined now. This relationship is similar to what we see between CentOS and Red Hat already.

CentOS takes RHEL and redistributes it for free. If enough customers of Red Hat were to opt for CentOS instead of RHEL, Red Hat’s revenues would suffer. Hence, Red Hat would cut expenses, including their R&D spending. This could lead to development quality and speed on new feature addition issues inside of RHEL. This, by implication would lead to the same issues within CentOS. As a result, fewer customers would use CentOS, thereby returning the balance between Red Hat and CentOS.

The same scenario applies to Red Hat & Oracle, at least in the near term. That’s why I have a real problem with the likes of Matt Assay crying bloody murder. DISCLAIMER: I don’t like Oracle’s business practices in general. I am incredibly happy that IBM knows how to contribute and participate within the Open Source community and that this is not a move you could really see IBM making. I’m just responding to the outcry from a business and customer standpoint.

From a customer standpoint:

1] If Oracle does a terrible job providing support for RHEL, customers will go (back to) Red Hat. There will be some customer pain while they migrate (back) to Red Hat support, but at least they aren’t locked in.

[2] If Oracle does a great job at it, customers win. (Unlikely if you ask me :-)

[2b] If Oracle wins a lot of (new) customers (from Red Hat) and inhibits the growth of Red Hat, this would cause Red Hat to spend even less on R&D than they do. Now, Oracle has a choice:

• [2b-1] Continue to base their support business on RHEL, even though the lowered R&D spending by Red Hat is resulting in a lower quality product with fewer new features. (Unlikely because this is a do nothing option and customers would lose.)
• [2b-2] Add more Oracle development resources to the Linux development community to make up for the reduced Red Hat R&D. (Likely because we have to assume Oracle will need to add some employees to the community ASAP to help support the unbreakable linux network. Hence, no net negative impact to customers.)
• [2b-3] Oracle acquires Red Hat. (Possible, but not without other large IT vendors jumping in to support Novell SUSE or Ubuntu to ensure that there is more than 1 large IT vendor endorsed/backed Linux distro. No net negative impact to customers.)
• [2b-4] Decide to fork. (Most unlikely, and this would have a negative impact to customers. But on the other hand, maybe it would increase choice for customers.)

[2c] Red Hat’s financial troubles make it an acquisition target for the likes of IBM, HP, Sun, Dell or even SAP. (Possible, but other major IT vendors would act to ensure that choice of Linux distros remains, so no net negative impact to customers).

[3] The Oracle competition just makes Red Hat work even harder at their core Linux offerings, which is where they’ve shown their ability to compete and win. (Most likely ;-)

Lots of roads this can take. Lots of possibilities, the majority of which don’t have a large negative impact on customers. It’ll be interesting to see this play out.

So, Oracle went ahead and announced support for Linux. Support for Red Hat Linux to be more precise. Instead of having their own version of Linux, it appears that Oracle is going to support RHEL better than Red Hat. Well…until Oracle decides to actually fork RHEL or the FUD brings Red Hat’s stock down low enough to be acquired.

Dave Dargo from Ingres, and ex-Oracle open source program office lead, has a very interesting take.

I can’t wait for Marc Fleury to pipe in, especially since he explained why this wouldn’t happen. But hey, it won’t be the first time Fleury et al. have tried to rewrite their past views/blog posts.

Update: Interesting reply from Red Hat about Oracle ULN here.

Infoworld has a biting review of Rob Levy’s (CTO at BEA) comments on open source a few weeks back.  I didn’t blog about this before because, well, I feel bad for BEA.  To go from having a big lead in the App Server space to, well, where they are now, must really suck.

And to be fair, I think that BEA’s “blended” strategy of allowing customers to use open source software and BEA’s commercial products together is a step in the right direction.  But if BEA really wanted to get behind OSS, they’d become more active in the Apache Geronimo community.  I am 99% sure that someone at Apache, IBM or internally would have brought this up to BEA execs already.

I won’t go into why it would be good for BEA to do this, as I’m still and IBM employee.  But if they’ve followed our revenue results, especially since we got into the OSS-based app server game with WAS CE, BEA would see it’s an easy decision to make.

There’s a lot of confusion about what Sun actually intends to do when they use the term “open source java”. We shouldn’t confuse the Java SE standard with the Java SE reference implementation.Check out Sun’s (new) Open Source Java page. If you look at the “OpenJDK” paragraph in the communities section, you’ll find that Sun is referring to open sourcing Sun’s implementation of Java SE. It doesn’t mention anything about governance of the Java standard, which is much more important to open source Java if you ask me. Now if you look at the “OpenJavaEE” paragraph, you’ll notice that they somehow tie GlassFish (Sun’s JEE implementation) to the governance of JEE:

“The community governs the development of specifications, technologies, and products that work together to provide enterprise-level solutions under the Java EE paradigm.”

Now call me crazy, but (1) doesn’t the JCP govern JEE and (2) wouldn’t a true open community like the Apache Geronimo community be a better place for the implementation and governance of JEE? Why is open source JEE so closely tied to Sun controlled efforts? I thought this was a community story.

Based on what Sun appears to be trying to pull with JEE, I have fairly large reservations about what open source Java SE (implementation and governance) will really look like in the end.

[UPDATED 2006-11-13: This post is getting a lot of hits, so I wanted to point readers to more recent posts on this topic]:

1. Sun Gets Press and Maintains Control
2. Matt Asay believes Sun gets it - But what is it?

DevX has an article on using WebSphere Application Server Community Edition (WAS CE) and AJAX together for rich, scalable web applications.

IBM developerWorks also had a previous article on using WAS CE and AJAX. You need an IBM ID (which is free) to view this aritcle. Yeah, I know it sucks to be forced to register to view content. But, getting an IBM ID isn’t a big deal, and it’s good to have to view any ‘protected’ articles at developerWorks. If you’re having trouble getting to the PDF, send me an email or post a comment and I’ll email it to you.

IBM is offering 30-days of free support for WAS CE.  This offer has been available for about 1 month now, and I’ve heard from the WAS CE team that the response has been quite strong.  And they expect even more adoption, since the free 30-day support has been extended to users who download WAS CE from Sourceforge/OSTG.  (Side note: WAS CE downloads via sf/OSTG are almost 10x more than through ibm.com)

This is a great idea that I’m sure is going to help customers choose WAS CE even if they haven’t had experience with IBM support in the past.  Removing barriers to adoption is a great benefit of the OSS movement.  It’s cool to see the WAS CE team applying those lessons to support also!

So go ahead, give WAS CE and the trial support a shot to get familiar with the quality of support you can expect from IBM for WAS CE.

According to this post on the Geronimo devlist, Geronimo 1.1.1 was certified on Azul appliances.

“The testing was performed on an Azul Compute Appliance Model 1920B (8 chip / 192 cores, 64GB memory) coupled with Azul VM release 2.3.1.2 (HotSpot™ Java™ 1.4.2_12), 64-bit Server mode. “

That’s quite the machine to get to play with! ;-)

Linux-watch reports that Novell & IBM have introduced the updated Integrated Stack for SUSE Linux Enterprise (ISSLE). The integrated stack includes:

• OS: SUSE SLES 10
• App Server: IBM WAS CE
• Database: IBM DB2 Express-C
• Hardware: IBM BladeCenter or IBM System x
• Make Windows & Linux play nice: Centeris Likewise Management Suite

The integrated stack is intended to help customers of all sizes lower their total cost of ownership when running File & Print servers within a Windows environment and provides a low-cost, rock solid application runtime environment.

It sounds like a great idea…too bad about the graphic that they’re using to describe the stack ;-)

Said the commercial software market of course ;-)

Matt Assay, I know you’re a big proponent of open source, as am I, but how about being a little more realistic here.

Red Hat has less than $300mil in total company revenues. While IBM doesn’t publish revenue for individual products, Gartner and IDC have both pegged WebSphere Application Server (WAS) revenue at well over $400mil and that was in 2005.

The revenue growth that the WebSphere Application Server (WAS) family has achieved over the past 5 years, again not public data, significantly outstripped the market growth rate for App Servers. A good proof point to this is to look at the WebSphere growth rate that IBM releases during their quarterly results calls. Yes, the growth rate is for a superset of WebSphere products, and not just for the WAS family. But, WAS is very likely the largest member of the WebSphere family from a revenue standpoint because it’s been around the longest and is the foundation for other products in the family. See the quarterly revenue growth for the WebSphere Portfolio below:

The growth of WAS revenue has been occurring in the face of much hyped competition from open source vendors like JBoss. But at the end of the day, JBoss is hoping to drive $20mil in revenue this year. The WAS team likely won’t be happy if their year on year revenue growth is 2-4x that number. (Although I don’t see internal revenue targets anymore).

Why has WAS succeeded in the face of open source competition? Well, because we decided to get into the open source application server game also. When we released WAS Community Edition (WAS CE) and threw our support behind Apache Geronimo, we gave our customers another choice in the open source app server arena. Having a product that was part of the WAS family and worked with other versions of the WAS Family and other IBM products was seen as a major benefit.

To say that open source will take over the software world assumes that enterprise vendors are either going to:
(a) do nothing about it
(b) fight open source

The other choice, (c), the one that the IBM WAS team took is to:

Enter the open source game and ensure that customers can choose from a spectrum of products, based on their specific needs and budgets, and that the OSS-based products increase the value of existing and future customer investments in the vendor’s commercial software.

Open source products and the open source business model is great, and I support both of them. But neither will replace the commercial software product or business model anytime soon or wholesale. I say this because commercial software vendors will get into the OSS game and the line between OSS and commercial will blur.