Dave Welch, CTO and Chief Evangelist
During Sunday’s Partner Event keynote, Larry Ellison said,”I think it’s enormously exciting that we can take this great company Sun and combine it with another great company Oracle and then merge those technologies and maybe do things neither company would be able to do by themselves. That’s our goal, to tackle problems even bigger, to integrate hardware and software, and deliver revolutionary systems.”
Monday Oracle President Safra Catz said: “That it turns out that us and all the other software vendors, were all sending you little pieces of technology all these years, and it was at your site that you had to make it all work together. And what we thought was that this really didn’t make sense. That long-term, companies like us had to take more and more of the responsibility of bringing you systems that work together.” She continued, “But many of you know there’s a reason we call this Oracle Open World, because we are just slavishly devoted to open standards. And if you want some of the pieces and some from others, that’s ok. We’re going to make that possible. But our hope is that little by little, you take many of the pieces from us.”
Yet it appears to me that Oracle has already started stepping away from openness, in the very moment when its hardware partners, OS partners, and customers rather need to be reassured as to Oracle’s commitment to a somewhat level playing field moving forward. I submit as evidence the “Database Smart Flash Cache” feature new in 11g R2.
11.2 New Features E10881-03 October 2009, p. 1-27: “New in Oracle Database 11g Release 2 (11.2), the Database Smart Flash Cache feature is a transparent extension of the database buffer cache using solid state device (SSD) technology. The SSD acts as a Level 2 cache to the (Level 1) SGA. Database Smart Flash Cache can greatly improve the performance of Oracle databases by reducing the amount of disk I/O at a much lower cost than adding an equivalent amount of RAM.”
Sounds pretty good. But get this restriction in the 11.2 Concepts Guide part # E10713-04 October 2009 p. 14-9) “Note: Database Smart Flash Cache is available only in Solaris and Oracle Enterprise Linux.” The 11.2 Admin Guide (part # E10595-06 October 2009 p. 6-21) says Database Smart Flash Cache is only supported on the Solaris or Oracle Enterprise Linux operating systems. This is confirmed by 11.2 Release Notes for Linux (part # E10838-03 October 2009 p. 2) and 11.2 Database Release Notes -03 p. 9. 11.2 Database Licensing Information (p. 1-4) says Database Smart Flash Cache is Enterprise Edition only and Solaris and Oracle Enterprise Linux only. The 11.2 Reference (E10820-03 October 2009 p. 1-49, 1-50) says nothing about OS restrictions on either of the new db_flash_cache_file and db_flash_cache_size parameters.
What technical justification could Oracle have to not make this feature available or not support it in the HP-UX, AIX, or Windows ports for example, let alone the Red Hat and SuSe Linux distributions? I’m feeling the need to reevaluate my enthusiasm for Oracle Enterprise Linux, for the first time since its announcement in 2006.
Mr. Ellison said, “There are some advantages of the single organization having control of the engineering of the hardware also engineering the software and engineering all the pieces to fit together well.”
It would appear that Oracle is already taking initiatives with respect to the Sun/Oracle announcement that give me pause. If we’re already seeing what may be a competition-motivated lock-out of a key feature like database smart flash cache today, what kinds of additional lock-outs may be coming once the hardware and DB software are subjected to co-engineering? I am not in favor of Oracle’s Sun acquisition.
Mr. Ellison continued in his Sunday keynote, “We are not selling the hardware business, no part of the hardware business are we selling. And we think Sparc is a fantastic technology, and with a little more investment it could be even better. You know the CMT line is fantastically innovate, and we think again by adding to the already spectacularly good engineering team we have around Sparc, we can increase our rate of innovation. Make the chip even more reliable while consuming less power. Make the chip even faster while consuming less power. So we intend to invest in Sparc.”
I’ve been dazzled by how elegantly Oracle has integrated opened new business lines beginning with the Oracle Application Server curve back at the turn of the decade. I had my doubts Oracle was going to make it with an app server, but they proved it to me, technically as well as with actual market share. Oracle has consistently impressed me with their integration of acquired companies both organizationally and technically. I’ve wondered since the Oracle/Sun announcement if Oracle was finally in over their head by attempting to get into hardware. This week I’m swayed over to believe that Oracle can probably pull off the integration and management of Sun’s hardware lines. Larry Ellison’s in charge and he’s a remarkably impressive visionary, strategist and tactician.
Mr. Ellison continued, “Solaris is unquestionably the number one enterprise operating system in the world in terms of linear scalability, in terms of reliability, in terms of features and functions. It is the leading operating system running the Oracle database. But it’s just the leading enterprise operating system period.”
Mr. Ellison continued, “And we’re very proud to be working with Sun to make sure that all the Oracle software runs better on Solaris, more reliably on Solaris, and faster on Solaris than it’s ever run before, and it runs faster there than it runs any place.”
So what are Oracle’s intentions with respect to Linux? I think Chris Josephes sheds useful light on the issue, especially as it relates to Oracle’s 2005 declaration of Solaris 10 as the preferred development platform for these 64 bit platforms: UltraSparc, AMD-64, and Intel-64.
Mr. Ellison continued, “We’re going to spend more, not less on MySQL. We think it’s a fantastic piece of technology, it’s extremely popular, it’s an open source product, and we’re going to increase our rate of contribution to that product.”
On Wednesday, Scott McNealy said: “The one I hear a lot about is MySQL. Open Source. Well, what’s going to happen? This is the volume, open source, low end database out there in the marketplace. Is it going to get clobbered? Well, I don’t think so. And again Larry has said, “It doesn’t compete with the Oracle database,” which is true, this competes with Microsoft.” Mr. McNealy put up a slide with this text: “’MySQL and Oracle don’t compete at all,’ said Mr. Ellison adding that he would not sell MySQL. Financial Times, 22 September 2009”.
So what about the whole engineering effort behind Oracle 10g with one of the prominent objectives being that it would compete with SQL Server’s relative ease of installation and operations? At the time we were led to understand that the single CD 20 minute 10g Database install and the improved manageability of 10g Database, was in part a response to Oracle’s concern about the hemorrhage of customers jumping ship for SQL Server. What about Oracle’s open admission in the fall of 2005 that Oracle Express was a direct answer to SQL Server 2005 Express? If MySQL and Oracle don’t compete at all, then one would have to talk themselves into the new concept that SQL Server and Oracle don’t compete at all. A MySQL customer today doesn’t have the obvious technical upgrade path to paid Oracle versions compared to an Oracle Express customer. What then could Oracle’s financial/business interest in continuing to promote MySQL? I’d like to hear an answer to that question, rather than Oracle continuing to attempt to convince us that they’re going to support and contribute to MySQL simply because they say they will. The existence of the Oracle employee assigned to MySQL documentation, that I met at Tuesday evening’s bloggers’ meeting, did more to convince me of the prospects of Oracle’s on-going support for MySQL, than all of these executive statements combined.
On Wednesday during his Exadata II comments, Mr. Ellison said, “The important thing here I think is our belief that eventually the virtual machine, the VM, and the operating system have to work exceptionally well together. That means they have to be engineered together.”
With respect to the last sentence, I’m afraid I’m not similarly disposed. VMware and the guest operating systems weren’t engineered together, although current versions were most definitely engineered with awareness of each other. On the other hand, this seems like a huge forward looking statement on Oracle’s part, and one that I believe sets up a straw man to fill market demand that doesn’t today exist. On Thursday while leaving the conference, I had an epiphany about my full-time experience in VMware’s OOW pavilion: not one person asked me about Oracle Virtual Machine or even mentioned it.