Julian Stuhler addresses a couple of fascinating acquisitions, both announced on the same day and each of which could have a significant impact on the world of DB2. Join Julian as he speculates on the facts, rumors and implications of these acquisitions.
Last month, I promised that I’d be talking in more detail about
the many performance enhancements in IBM DB2 for z/OS Version 10. I’m going to
keep you hanging on for another month, as some interesting IBM lab measurements
are just beginning to trickle out and I want to be able to cover them as fully
as possible. Instead, I want to address a couple of fascinating acquisitions
that were both announced on the same day, each of which could have a
significant impact on the world of DB2.
Xkoto and Teradata
Firstly, Teradata Corporation
acquired Xkoto on 12th May 2010.
Xkoto was the creator of some very interesting clustering technology dubbed
GRIDSCALE, which was designed to improve database availability and scalability
with minimal impact to the application. If that sounds familiar to you as an
IBM DB2 person, you’ve probably been reading about IBM’s pureScale
feature for the DB2 for LUW product, which is aimed pretty much at exactly the
There are a number of important differences between the
GRIDSCALE and pureScale technologies (including the database systems supported,
licensing/hardware implications, TCO, disaster recovery capabilities and many,
many more) and there have been plenty of lively discussions on the relative
merits of each. However, all of that suddenly became irrelevant on 12th
May because at the same time as Teradata announced the Xkoto acquisition,
GRIDSCALE was withdrawn from sale. Existing customers will of course continue
to be supported, but it is no longer possible to purchase new licenses.
It’s hard to speculate exactly what Teradata will do with the
technology they have acquired. GRIDSCALE was pretty much the only product
offered by Xkoto and there isn’t an obvious natural fit between that product
and the rest of Teradata’s existing portfolio as it only runs on DB2 for LUW
and Microsoft SQL Server. However, one thing is certain: right now, if you’re
in the market for a solution that will increase the resilience and scalability
of your DB2 application, IBM’s pureScale feature is the best (and only) game in
PureScale has had in interesting history. Its development was
kept a closely-guarded secret until October 2009, when it was announced with
much fanfare at both IDUG Europe and the global IOD conference in Las Vegas. It
quietly became Generally Available in December 2009 but relatively little has
been heard about it since.
Appearances can be deceptive. Potential customers are running
Proof of Concept projects, business partners such as Triton Consulting are busy
building their skills so we’re able to support those customers, and behind the
scenes IBM is still very busy working to further improve the technology and
plug some of the gaps in the initial release. I’m expecting to see some more
major announcements in this area over the new few months, which will once again
bring pureScale into the limelight.
In the meantime, SAP has also
been working closely with IBM on making their application platform run well on
a pureScale platform (in fact, they have been working with IBM for nearly two
years, almost since pureScale’s inception). Although SAP has not yet formally
announced pureScale support for their NetWeaver platform, they have begun to
talk openly about it at conferences such as the recent IOD Europe event in
Rome. With many SAP customers running high-volume, mission-critical
applications using IBM DB2 as the back-end database, pureScale is a natural fit
and I’m expecting a significant proportion of the early pureScale customers to
be running a SAP workload.
Sybase and SAP
Which brings me neatly to the second of the two announcements
made on 12th May: SAP’s intention to acquire Sybase for the princely sum of $5.8 billion. With
IBM DB2, Oracle Database and Microsoft SQL Server now dominating the database
market, Sybase is generally regarded to have lost its way somewhat in recent
years, despite having a mature database platform and some great technology and
partnerships in specific areas such as mobile synchronisation.
The official SAP press release majors on the enhanced mobile and
in-memory computing technologies that Sybase will bring to SAP, and this is
indeed very much in line with SAP’s technical vision. However, the real question for us is, “W this announcement will alter DB2’s position as SAP’s
preferred database for customer implementations?” Surely, the fact that SAP has
just acquired its own highly capable database means that DB2 will be sidelined.
Well, that’s far from a foregone conclusion. Right now, Sybase
isn’t even one of the supported platforms that SAP will run on. Both SAP and
Sybase run their own internal SAP systems on…DB2! Look at any release of DB2
on LUW or z/OS during the past few years and a large proportion of the new
features are designed specifically with SAP in mind (of course, they are usually
of benefit to other ERP and bespoke dynamic SQL applications as well). Since
SAP and IBM formed their strategic partnership in response to Oracle’s own
acquisitions several years ago, hundreds of man-years of development have gone
into enhancing both products to work more efficiently together. That effort has
paid significant dividends, with DB2 being very favourably represented in SAP’s
The existing partnership makes good strategic sense for both SAP
and IBM. From SAP’s perspective, it has access to a highly scalable, resilient,
mature database that is able to perform very well with its NetWeaver
architecture. New features such as pureScale only add to that story. SAP’s
abstraction of the underlying database (from both an application and an ongoing
administrative point of view) makes it relatively painless for SAP customers to
switch databases (over 100 SAP/Oracle customers moved to DB2 during the past 12
months). For those SAP customers also running bespoke applications on Oracle,
the Oracle compatibility features in DB2 for LUW 9.7 make it much easier to
port those to DB2 as well. The fewer of its customers that are dependent on its
number one competitor’s technology, the happier SAP will be.
IBM has similarly compelling reasons to continue to partnership.
A significant proportion of the net new DB2 licence sales on both LUW and z/OS
platforms have come from SAP customers in recent years, which is why SAP-friendly
features always feature so prominently in new DB2 releases.
There is of course always a risk that SAP will decide to adopt
Sybase as its strategic database at some point in the future. The very same
database abstraction that makes it less painful for customers to migrate from
SAP/Oracle to SAP/DB2 also makes it easier for SAP to move to support a new
database such as Sybase. There are also some obvious financial attractions for
SAP being able to offer a complete application stack with the database
With a fantastic sense of timing, just a week after the SAP/Sybase
announcement, IBM went public with a new DB2 feature to allow Sybase
applications to be ported to DB2 with minimal changes. This new SQL
Skin feature (jointly developed with ANTs
Software, Inc.) allows Sybase ASE applications to run natively against DB2
at a tiny fraction of the cost of a traditional migration. The final irony is
that the ANTs ACS technology used in the SQL Skin feature is designed to be
able to work against any database, and could eventually be deployed to allow
Oracle applications (including SAP) to run against Sybase!
So we’re left with a potential acquisition that has several
months of regulatory approvals to go through, and a whole lot of speculation
and theories in the meantime. Personally, I believe the most likely outcome is
for Sybase to be adopted for low-end SAP solutions, replacing SAP’s existing
(and under-developed) MaxDB database. For more demanding environments, it will
take years of development effort for SAP to be able to enhance the Sysbase code
to the extent that it can match DB2 as a scalable, reliable, cost efficient
NetWeaver database engine. I’m not convinced that the potential gains outweigh
such a monumental effort.
However it turns out, the next few years promise to be very
interesting ones in the ERP database market.