Suppose someone high up in your organization announces that your data center choice between IBM's AIX on Power offerings and Sun's Solaris on SPARC line is more of a business decision than a technology decision - thus both abrograting and announcing the intended decision. What do you do?
I think your review should focus on four distinct sets of issues:
- what application availability tells you about the decision;
- what the two technologies imply about future business strategies;
- how costs compare; and,
- what the relative threatscapes look like.
One note: when I started this series of five blogs I didn't think they would be hard to write, but as it turns out the first two have been, and I doubt that will change for the other three. What's going on is that I've worked with people from both cultures, used both product sets, and consequently have some difficulty seeing the good things on IBM's side. Think of this, please, not as an admission of bias (perish the thought!) but as a challenge to readers: catch me where I'm wrong..
Anyway, to get through this, lets look at application availability today, at technology tomorrow, and review various forms of cost next week.
Notice that if you're interested in helping shape larger scale corporate business decisions on long term platform partners you're really not going to care much about either company's claims to run most Linux software or about the cultural implications of either architecture's presence on Sourceforge. What you are going to care about are first your own applications - do they work? are they supported? - and secondly what third party vendor commitment to either "platform" can tell you about how people are likely to view your decision five years from now.
Here's how IBM describes their ISV community:
Independent software vendors recognize AIX as a premier UNIX operating system. Over 3000 ISVs now support over 8000 applications on AIX 5L. Support for your core application is most likely available today. For more unique applications, IBM offers a rapid process for porting applications to the latest AIX platform.
Click on that link and what you get is their "Business Partner Application Showcase". This is sorted by industry and selecting "view all" for the first one shown, banking, gets you the first ten of a claimed 1,194 search results.
Four of the first ten are in English, three are in Asian languages about which I have no clues (except that the technical environments are listed in English), one is in Italian, and two are in German.
Although AIX 5L V5.3 is not mentioned in any of the ten listings, five mention AIX or AIX 5L V5.1 as run time environments. Two require Windows, one requires Red Hat Linux, one OS/400, and one zOS.
Google returned no hits outside IBM's promotional environment for two of the seven products listed in non Asian languages - and the references to "OS/400" and "Windows 2000" in two others gives reason to suspect that these listings are well past their best before dates. AIX 5L V5.1's nominal end of service date was Dec 31, 2005.
One firm, India's 3i Infotech, has two English language listings among these first ten - and both apparently support AIX. In total there are nine listings for 3i's "Kastle" applications - but four of those are duplicates and one of those is listed a third time under the name: "Triton - Universal Lending Solution".
Quickly and repeatedly clicking on "Next" caused IBM's site to lose the page and item counters, but it eventually stablized on a list of ten starting with "Paxus RFAF Real-time Fraud Application Feed". That page contains one AIX listing (For 5L V5.2), four for Windows, one for zVSE, and four for zOS.
Overall, therefore, I think we're entitled to suspect that no more than half the 1,194 applications returned by searching IBM's partner showcase for banking related AIX applications actually run on AIX, that a majority of these are not 5L V5.3 certified, and that a significant number of the actual AIX entries are duplicated.
In contrast all of the first ten applications (including 3i Infotech's "Bank Alert") on Sun's list of 4,349 Solaris 10/SPARC products exist and run on the current ultraSPARC/Solaris 10 product set.
Sun links the user directly to the software partner's home page while IBM links only to IBM's own listing for the product. Sun's stuff is, therefore, as up to date as the partner's own website, where IBM's approach leaves you to first find, and then grope around in, the vendor's website looking for a current product write-up.
As a check on currency, take a look at what happens when you do a google search on "AIX 5L V5.3 "the leader in" announces -site:ibm.com". The first page includes one hit, ironically Sun's late 2006 Netbeans availability announcement for AIX. In contrast that same search, modified for Solaris 10 yeilds six product announcements, all recent.
So what's the bottom line? Well, we're not going to get definitive numbers on the ISV universe without spending a lot more time on this, but it's probably fair to suggest that the Solaris application base is counted in the thousands and growing where the AIX base is counted in the hundreds.
So unless somebody wants to take the time to evaluate every listing from both companies what this comes down to is a simple bottom line bet: if you pre-commit to Solaris on SPARC and then start hunting for just the right application, your chances of success are on the order of ten times better than if you pre-commit to AIX on Power.
Or, if you prefer to see the glass as half empty: the chances that the lack of a critical application forces you into a risky and expensive development or porting project are at least ten times greater with AIX/Power than with Solaris/SPARC.
See: Part 2