IBM Gets Feisty — Mobilizes Analytics for Oracle Battle

In July 2009, IBM announced the Smart Analytics System 7600, a workload-optimized, pre-integrated bundle of hardware and software targeted at the business analytics market. Included in that package are an IBM POWER 550 running AIX, storage, plus InfoSphere Warehouse Enterprise Edition (which consists of DB2, Warehouse design and management tools + Cubing, Data Mining and Text Analytics services), and Cognos 8 Business Intelligence, configured and tuned, and “health check” features. Accommodations are made if the customer already has licensed some of the software and wants to use it on the platform; in this sense, the software is described as “optional.” This month, IBM broadened the story and upped the ante, making Smart Analytics System a key weapon in its widening battle with Oracle.

This post is a slightly updated version of a piece that appeared in the PUND-IT newsletter.

In the year since the first announcement, IBM’s cadence around workload optimization and workload optimized systems has continued with software updates, the closing of the SPSS acquisition, and messaging about marketing, sales and partner programs to carry this key strategic thrust forward. As the second quarter of 2010 began, IBM’s Systems and Technology Group (STG) rolled another set of assets into place, announcing new workload-optimized systems for IBM’s System x (x86) and System z (mainframe) platforms, an aggressive campaign targeted at Oracle migration, and a $500 million war chest for financing Sun Microsystems partners who want to resell Big Blue’s systems.

However, the event was thin by comparison to earlier ones. The Services and Research teams now mobilized into the field to provide leading-edge implementations for customers were missing from this event, though they have played a role in other similar rollouts. And while the availability of Smart Analytics Systems for System z, System x and POWER-based operating environments (zOS, Linux, and AIX) grows the prospect audience, little new functionality to extend the design was in evidence. Software bundled into the offerings is the bare minimum, and many key components continue to be described as “complementary,” which suggests that not only costs but deployment and integration efforts will be higher when they are included. Some additional hardware opportunities are yet to be included – for example, ProtecTIER hardware-based data deduplication, which could add as much to the value picture as IBM’s effective data compression features in DB2. And IBM continues to miss the opportunity to ride the collaboration wave in BI by leveraging its Lotus assets; Lotus has been absent from every BI and analytics-related briefing I’ve seen in the past year. [correction 4/16 – I’ve been reminded that I was briefed under NDA last September about Cognos-Lotus integration plans. Nothing for public consumption yet, though.]

IBM clearly has a multi-front battle in its plans. From the outset of the Business Analytics and Optimization (BAO) organization’s formation under Ambuj Goyal, then in SWG, it was clear that IBM was marshaling assets for maximum impact. Leveraging its vertical integration across hardware and software is one key weapon, and Goyal’s promotion from his software role to general manager of STG’s Development and Manufacturing team reinforces that theme. He kicked off this event at IBM’s Almaden Research Lab, where Ted Codd did much of the early theoretical work on the RDBMS model that dominates today’s database market. At the event, Goyal laid out the aspirations of the campaign, then handed the microphone to his successor, Arvind Krishna, now general manager, Information Management in SWG, to introduce the two new Smart Systems offerings that begin to deliver on the vision laid out last year.

IBM rolled out the Smart Analytics System 9600 atop System z and the 5600 for System x. Like the POWER-based 7600, the systems integrate server hardware, storage, and software. (On the z, Cognos runs on Linux; its native z implementation has just entered beta.) These key pieces are pre-integrated by IBM and sold together and can be rapidly functional on customer premises as a result, resulting in lower acquisition and operation costs. The advanced workload management, modular storage and health-check services make for a powerful value proposition.

However, the slide headline: “EVERYTHING you need for Business Analytics” [emphasis IBM’s] was belied by the number of elements that are “complementary.” The 9600 comes with hardware-assisted data compression and advanced encryption, and boasts online database reorg for migrating DB2 table spaces to the appropriate drive tier, in addition to online backup. Server and disks both have the health check “call-home” features to help ensure that failures are averted before they happen. So IBM has taken a good step in the right direction, outpacing competitive offerings with its integration efforts. But on the software side, DB2 Connect, InfoSphere Information Server and Replication Server, and Tivoli OMEGAMON for DB2 are all complementary, even though it’s hard to populate and optimize a warehouse without them, and SPSS was not mentioned at all. IBM’s integration of SPSS got off to a fast start last year because of the prior relationship the two firms had, and IBM shared a number of planned next steps with me in a discussion of this year’s roadmap. But it’s not being bound into the story yet; the firm’s typical reluctance to pre-announce and pre-market offerings is holding back a story that has the potential to be the most complete in the industry, making the “appliance” message heard elsewhere sound thin by comparison.

Taking On Oracle

In keeping with the hardware-led message, much was made of the full-system performance profiles, focusing on messages like “50% less floor space,” “70-90% lower energy consumption” and “90% server utilization.” IBM took Oracle on pointedly here, showing direct benchmark comparisons demonstrating the ability to use fewer cores to achieve better SAP-on-DB2 results. A customer panel drove this home; the IBM partnership with SAP was cited as a key factor in customers’ decision to switch from Oracle. (The customer names, although shared with the analysts, are under non-disclosure due to legal conventions.) One, with 8,000 users, closely compared SAP’s own MaxDB, Oracle and DB2, and picked IBM based on cost and performance. The others agreed, and cited IBM’s tools for SAP-DB2 as improving administration. All cited the stability and ease of the migration effort itself. Two said they achieved 40-50% savings on storage alone due to data compression with DB2.

One customer involved in a major construction project announced that it “did our migration on a gas-powered generator.” The company finished its Oracle-to-DB2 migration a day earlier than planned (similar stories were heard from the others) and achieved across-the-board performance gains. One cited an “18% performance improvement out of the box.” IDC’s Carl Olofson and I discussed some of the technical reasons for this, and Carl cited the fact that “IBM substantially rewrote their buffer and lock management in DB2 LUW (Linux, Unix and Windows), and now they are reaping the rewards.”

Much was made of the simplicity of software migration from Oracle to DB2, which substantially contradicts conventional migration wisdom, and another set of panelists went into details of this. A demo station featured the technical details for those of us wanting a deeper dive, and an IBM engineer and I discussed some of the extensive additions IBM made to the EnterpriseDB software it licensed last year. WebLogic to WebSphere migration was also discussed, and IBM appears to be ramping up a strong competitive effort here in the middleware tier as well; I expect to hear more about this next month at IBM’s IMPACT show. Licensing issues are also a key factor— multi-core systems have a huge impact on database licensing costs, and several customers cited the fact that IBM’s licensing was far less costly than Oracle’s as a key factor in their migration decision.

Putting the Ecosystem to Work

It’s easy to be critical, as I am above, of the pace and breadth of IBM’s bundling of hardware, software, services and research efforts into its competitive offerings. But it is still a formidable undertaking, in an organization as large and complex as IBM. The progress has been steady and measured. On another front, IBM has been investing in the marketing and sales efforts to support these solutions. Literally dozens of pieces of collateral were made available to analysts on USB disks, and sales training and price list changes have been underway for months. And then there is the partner network. It’s a crown jewel for IBM, and new programs, incentives, and support systems are being made available to drive these offerings into the market.

Field numbers and readiness count for a great deal in a global business. IBM announced that it is earmarking $500 million to entice Sun Business Partners to resell IBM systems. It’s easy to underestimate the importance of this, but it’s a huge play. “The gloves are off,” Forrester’s John Rymer tweeted, and he’s absolutely right. It’s a strong response to Oracle’s decision to take the largest (over 4000) Sun accounts in-house recently; clearly IBM hopes to capitalize on the disaffection this will create. Another point of comparison: IBM’s  investment is double the size of the recent HP-Microsoft joint agreement. It ought to buy a lot of points of presence for IBM in the year or two ahead. IBM also said it will make tools like its profit optimizer application and working capital options available to Sun partners.

Wrapping it Up – Customers Will Win

IBM also announced the new pureScale Application System, combining POWER7-based servers with WebSphere Application Server (WAS) and DB2 pureScale software to target transactional workloads. “Scale-up, Scale-out and Scale-within” is the message here. The platform can scale from 8 to 8,192 cores, an astonishing range, deployed in a flexible form factor that will allow customers to start small and grow. Clearly IBM is ceding nothing to the opposition.

Solid-state disk (SSD) technology was another option on these new systems, as it is for Teradata, Oracle and others, confirming that SSD’s steadily dropping price is making it increasingly attractive as a design point for performance optimization.

Clearly, this is the beginning of the transformation to a battle based on full-stack offerings, and the customer wins if vendors compete on ease. The battle raises intriguing challenges for both vendors’ partners. For example, Oracle will certainly optimize its own stack first, and then its partners’. But its partnering muscles will need training; IBM has been partnering with all comers far longer in more spaces.

Who’s left out of the full stack story IBM is painting about what makes a true strategic partner? HP: no software across most of the stack. Microsoft: no hardware and minimal services. And for Oracle, even with Sun hardware and an increasing services presence, controlling it all will be a big challenge that will take time. An even bigger challenge: extending its services expertise to take advantage of most, if not all, potential opportunities. Field and partner training are key. The battles are not really started yet; only the first skirmishes have been seen, but all the vendors are moving their resources into position. 2010 and especially 2011 will see pitched battles for key account ownership. It’s a great year for the buyer.

Disclosures: IBM, Oracle, Teradata and HP are clients of IT Market Strategy

About Merv Adrian
Gartner Research VP, technology analyst and consultant, 30 years of industry experience, covering software mostly, hardware sometimes.

5 Responses to IBM Gets Feisty — Mobilizes Analytics for Oracle Battle

  1. Pingback: IBM Gets Feisty — Mobilizes Analytics for Oracle Battle « Judith Hurwitz's Cloud-Centric Weblog

  2. Merv, overall good blog but as a member of IBM analyst relations its my duty to point out one inaccuracy. Regarding collaboration and BI you point out “And IBM continues to miss the opportunity to ride the collaboration wave in BI by leveraging its Lotus assets; Lotus has been absent from every BI and analytics-related briefing I’ve seen in the past year.”

    Current shipping product IBM Cognos 8 BI v8.4 does have a level of integration with Lotus Sametime and Lotus Connections for ‘collaborative BI’ today (but to your point maybe we don’t talk about it much). I would point you to the NDA-level materials we presented at the Sept 09 Cognos Analyst Summit for other product plans we outlined in the collaborative-BI arena. Hopefully that will refresh your memory. Take care.

  3. Merv Adrian says:

    mea culpa You’re absolutely right. This is one of those “what did he know, and when did he know he knew it?” questions, I guess. Obviously I did a good job not talking about an NDA conversation. Thanks for the input and the specific pointer; the slides were indeed there. I’m going to make an edit to the document for future readers who don’t read comments. And of course, I’m looking forward to seeing IBM talk about it – in a disclosable way. ;-)

  4. Merv

    We have worked for the last 5 years on integrating the world of BI with Lotus Collaboration starting first with Cognos and then opening up th technology – take a look at http://www.sunandson.com and then look at the white papers under the rsoures tab for fuller details of the work we have done with the IBM CIO office in this area and the IBM Cloud CTO

  5. Merv Adrian says:

    Thanks – I’ll look forward to checking it out.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Follow

Get every new post delivered to your Inbox.

Join 17,210 other followers

%d bloggers like this: