News Review: Oracle License Clustering from Compaq
Compaq License Tru64 Clustering Technology to Oracle. In return for an undisclosed sum, Oracle gain ready-made clustering for Oracle 9i. But does this signal a change of direction for Compaq's Unix Strategy ?
Compaq and Oracle have announced a three-year development and marketing agreement covering Oracle 9i and TRU64 (UNIX) Clustering.
The basis of this is that Oracle gain a license to use the Compaq TRU64 Clustering Technology, which will become embedded into the core of the Oracle 9i clustering.
What this effectively means is that customers who require a Clusterable version of Oracle 9i will receive the Compaq technology as an integral part of the database. This is great news for Oracle, who have a habit of trying to take over control of hardware, in areas which were previously the responsibility of the Operating System.
Examples of this include the fact that Oracle now manages "out of the database" objects (such as video and bitmap files). A few years ago, they also propounded the famous "Raw Iron" strategy, whereby they would ship Oracle on a Sun hardware platform with a minimalist stripped-down Solaris kernel, suitable for just running Oracle, and nothing else.
But is this announcement just another example of Oracle Hegemony ?
Why Did Oracle chose Compaq ?
With Oracle 9i, there is a new Architecture called "Cache Fusion", which is intended to allow multiple machines to share Cache between them. This effectively means that full Live-Live clustering can be achieved, with the potential of massive scalability by "Scaling Out" to possibly dozens of Unix boxes to run one Database.
The unique aspect of this is that Oracle-based clustering becomes part of the Database itself, and is not relegated to an Operating System Function. This means that Oracle have greater control over the behaviour of the cluster, and can guarantee better levels of availability and performance.
However, Clustering is difficult to engineer, and the Compaq (ex-Digital) engineering is probably the best in the industry. So instead of writing it themselves - as they had tried in version 8 - Oracle licensed the technology from Compaq.
What does Compaq Gain ?
No doubt Compaq were very pleased to receive cash from Oracle. But was this just a money transaction ?
Some people think that this signals a change of heart within Compaq, and a recognition that TRU64 is potentially a powerful challange to AIX, Solaris and HP-UX in the commercial Unix arena. Up til now, Compaq has not really marketed Unix in the way it could - preferring to rely on it's relationship with Microsoft to sell it's NT-based servers. But this could be changing.
To quote one analyst, TruCluster is arguably the best clustering environment in the Unix world and the fact that Compaq would give away some of the family jewels ... indicates how serious Compaq is about advancing the use of clustering technology
Market Breaking ?
So are Compaq intending to take on Sun and others in the high-end Unix battleground ?
At the same time as selling Microsoft Windows 2000 solutions ?
It would take a lot of effort for any company to go to market with TWO messages - UNIX and WINDOWS2000. But others have done it, maybe Compaq can too. They certainly have the high-end equipment. The Alpha chip is still capable of taking on the Ultra-Sparc and PA-Risc chips, and the ex-Digital Unix is stable and robust.
Not only that, but the new high-end Compaq GS160 and GS320 multi-way machines could be capable of fighting off ES10000 or HPSuperDome machines for shares in the datacenters of Financial Institutions and Telcos.
There would be a lot of work required to convince Independent Software Vendors to start treating TRU64 as a tier-one Port again. Many of them abandoned the old OSF/1 platform some time ago.
But if every commission-hungry Oracle salesman who wants to close a high-end Oracle 9 deal starts recommending Compaq as the hardware supplier, then Sun will have to start looking to it's laurels.
Some Analysts suggest that this deal could give Compaq a six- to nine-month lead over its competitors in selling high-end clusterable Unix machines for supporting Oracle. Other sources suggest that the difficulty of "re-engineering" the Compaq clustering so that it can be used to control, say Sun, HP or IBM Unix servers, could take far longer.
Either way, Compaq appear to have been given a significant amount of money, and have a window of opportunity to establish their credentials in the massive high-end clusterable Unix market.
Personally, I hope they will succeed, but prior experience with Compaq suggest that the PC-makers from Texas still do not realise the potential of the mid-range technology which they bought from the East-Coast Digital Equipment Corp.
Reviewed by Dennis Adams in March 2001