IBM’s System i Priorities for 2007
January 8, 2007 Timothy Prickett Morgan
As the holiday season approached last December, two key people at IBM‘s System i division took some time out of their schedules to talk about the technology and marketing areas that they would focus on in 2007 with the i5/OS platform. Many of the priorities that Big Blue has for the platform have been talked about in 2006, such as Voice Over IP telephony and the PHP language running on the platform. But these key IBMers did offer some more insight as well as an initial reaction to my own suggestion that IBM fully embrace user-based capacity pricing for the System i5 line.
Normally at this time of the year–either late in one year or early in the next one–IBM is somewhat eager to talk, in a very general manner, about the upcoming hardware and software technology that will be rolled out in the next year or so. This time around, IBM is holding its cards a little closer to the vest. It is hard to say if this relative quiet means IBM is not, as some rumors have suggested, planning to make any substantial System i5 or i5/OS announcements in 2007. As I pointed out in stories in November and December last year, I don’t think IBM can wait until late 2007 or early 2008 to revamp the System i5 product line to make it more competitive in the small and medium business market. What I don’t know is when Power6 processors will make it into the System i5 line, thereby providing the normal opportunity to revamp and reprice the product line.
As it stands, Power6 looks likely to come out in the System p5 AIX and Linux server line before the end of 2007, and possibly in the System i5 line in late 2007 or early 2008. And the next release of i5/OS, which is the V5R5 release, is rumored to be coming out in early 2007–in time for the COMMON trade show and expo, but then again there are other rumors going around that V5R5 will be held up until later in 2007 or even possibly to coincide with the Power6-based server launch–whenever that happens.
When I spoke to Jim Herring, director of System i product management and business operations (he does the technology side), and Ian Jarman, product manager of the System i product line (he does the marketing side), in late December, the timing of the Power6-based servers and future i5/OS V5R5 release was not made any more clear. I am not sure why, but I think one reason is that the Power6 chip is about six to nine months behind schedule–partly because of delays in the Power5+ chip, which was about six months behind schedule when it was fully released across the System i and System p product lines. (Saying that anything is late or early in the computer business is a tricky art, since no chip maker ever publishes precise delivery schedules, and even if they do hint at timetables, they do so with very wide error bars.)
I also think–and this is based on a gut reaction, not any data–that the System i division is not exactly happy about getting Power6 chips well behind the System p line, which happened with the Power4 chips back in the early 2000s. While the iSeries got the Power5 chips first, the System p line got the Power5+ chips first, and that means it is System p’s turn to go second. But with the AIX and Linux Power-based server line generating around three times as much revenue as the System i5 line–compared to about equal sales in the late 1990s–it is not hard to see who is going to get to go first from now on. Unless, of course, IBM realizes that if it revamps the System i5 line to the System i6 line with aggressive pricing, it can tap into a much more vast and underserved market and boost revenues, getting a little bit closer to the revenue stream it is deriving on AIX and Linux machines using Power processors.
Hope springs eternal, and to be honest, this kind of revamping has happened in the AS/400-iSeries-System i line before, specifically in 1995 and 2003, and to a lesser extent in 1994, 2000, and 2005.
In the meantime, the two things that Herring and Jarman are excited about are the prospects for Voice Over IP telephony being deployed in conjunction with 3Com on Linux partitions on the System i and integrated with back office and front office applications. “The partnership with 3Com has turned out really well,” said Herring. “Most of us who played with early VOIP were not impressed, and this stuff is clear as a bell. People all over the place are looking at this 3Com solution.” Herring said that IBM’s own Lotus middleware software will be integrated with the 3Com VCX suite, which is the IP telephony product. Herring says that this “click to call” integration will also be available with an unnamed middleware software supplier partner of 3Com’s very soon. “I think you will see an expansion of this product beyond PBX replacement to application integration.”
PHP, which was brought to the System i platform through a partnership with Zend Technologies, has been downloaded close to 5,000 times by iSeries and System i5 shops running i5/OS V5R4, the current operating system release. In mid-December, Zend announced that it had put out a beta of Zend Core for i5/OS that will also integrate with OS/400 V5R3, the prior release, which has a much larger installed base. “I think this will open a lot more doors for PHP at a lot more customers, and I think you will see us take a more active role in promoting PHP on the System i,” said Herring. I asked if IBM was looking to partner with SugarCRM, the maker of the popular open source customer relationship management suite that is written in PHP. “We like Sugar,” is all that Herring would say, and you could hear him smile over the phone. And well IBM should. SugarCRM already has over 1,000 paying customers and over 1 million downloads.
Another area of interest in 2007 for the System i line is business intelligence. “We have a lot of customers who have experimented and deployed data marts and data warehouses on other architectures, and now they are telling us that they want to do decision support and business intelligence on the System i,” said Jarman. Data warehousing was the darling killer app in the data center in the late 1990s, but it never took the AS/400 and iSeries market by storm the same way the Unix and then Windows markets saw a big boost. Part of that has to do with cost. Microsoft embedded a freebie Online Analytical Processing (OLAP) engine in SQL Server 2000, and that was by some accounts the main reason that a lot of customers used SQL Server; transaction processing was not the main reason at a lot of shops. Exactly how IBM will capitalize on this desire to move data warehousing and related business intelligence processing back to the System i platform is unclear.
Of course, with the spate of natural disasters, which affect all companies, and the issues of compliance, which affect public companies, you can expect IBM to continue to work with its high availability and disaster recovery partners–which means DataMirror, Lakeview Technology, Maximum Availability, and Vision Solutions after two years of mergers in this business–to push high availability clustering and disaster recovery solutions and therefore drive more System i sales.
Jarman says that in addition to these broader areas, IBM will be focusing on building out a “value network” around smaller independent software vendors in specific vertical markets to drive sales. He was not at liberty to provide details on this.
Neither Herring nor Jarman wanted to talk all that much about one of my favorite topics since October: user-based pricing for the System i line. Both said that they had read the analysis I had put together with great interest, but did not endorse my way of thinking very enthusiastically. “The pilot that we kicked off back in October with the Solution Edition–and I would stress that it is a pilot–is really helping us to tune our ideas about user-based pricing,” explained Herring. Jarman explained that IBM sees user-based pricing as a means of competing mostly at the low end, and mostly against Windows, and the impression I got was that we should not expect a broad and certainly not a complete System i5 line with user-based pricing. “Competing with Windows is paramount to us,” said Jarman, “But in the Unix market, the dominant pricing method is still per-processor pricing.” The implication is that IBM still thinks of its competition for anything larger than a single-core i5 520, and maybe a dual-core 520 box, as a Unix server, which means the i5 should be priced accordingly using processor-based pricing.
I’ll give you just one guess to come up with what I think about that. While Herring and Jarman know the server market as well as any of their counterparts in the server market, I happen to disagree strongly with this assessment.
Still, there is hope. Jarman conceded that there is a fundamental shift in server pricing underway, with virtualization and other technologies coming into play, and Herring said that user-based pricing “lets IBM appeal to a broader set of customers than the i5 Value Edition and Express Edition machines can.” And then he added the most important idea: “We are quite comfortable that it is a practical system to have a user-based pricing structure that will be acceptable.” Acceptable to customers, resellers, and IBM, presumably.
Let’s hope IBM does more than a pilot in 2007.