Tuesday, December 06, 2005

Weaving knots rather than nets

SAP's Netweaver initiative is an astute move to try and define and so to control a standard for applications in large enterprises. It will certainly present advantages to existing committed SAP customers, who will be able to interact more easily with some non SAP applications. However there are several drawbacks. Firstly, NetWeaver's integration is skin-deep. If you are on an SAP screen you can potentially branch out to a non-SAP routine e.g. a specialist payroll calculation routine. However anything that requires the integration of data between SAP and a non-SAP system is no better off than they are today i.e. customers are still into coding. Since the higher value levels of integration will usually involve not just portal-like "put in on the same screen" integration but actually dealing with business meaning of data, this will limit the use in reality. For example Netweaver would allow you to drop out of an SAP process into a supplier system, but does not help you deal with the issue that your set of product codes, or your general ledger structure, are different from that of your suppliers. For meaningful integration you need to resolve the business meaning or semantics.

What Netweaver certainly does is to declare war on several industry players who were previously either neutral to SAP or indeed active partners. IBM is the most obvious example. IBM has a massive services arm that does huge business implementing SAP, and IBM has decided to stay out of the applications business, so relations between the firms were good. Netweaver directly attacks IBM's core websphere middleware, and so now IBM's software group is in direct competition with SAP. The same would go for the EAI and ETL vendors (e.g. Tibco) and Microsoft, who have their own middleware stack. Oracle competes here too, but of course Oracle was already the most direct competitor to SAP. SAP may not care about the EAI world, but IBM especially is a big target to take on. The reason that SAP is prepared to take this risk is that the reward is so great: Microsoft showed the power that can be exerted by controlling the critical standard, in their case Windows.

Most large corporations have multiple middleware stacks within their organization (SAP, but also IBM Websphere, Microsoft and probably Oracle as well), so the key issue for them is how to deal in a neutral way across these, rather than how to rip all but one of these out. This is where the Netweaver strategy may ultimately fail, since the sheer cost of ripping out an existing well-established software infrastructure is gigantic, and that assumes that corporations donÂ’t care about the lock-in that would give SAP. Some won't care about this, but many will. However the practical problem is the sheer scale of core infrastructure that would have to ripped out, and yet without data and business semantic integration, the benefits of doing such a thing would be very limited. Oracle and SAP are both giants locked in a battle to gain a larger and larger footprint in the enterprise, yet both are too well-entrenched to ultimately destroy the other. SAP has no database, for example, an Achilles heel for it, and SAP grieves every time they win an application account from Oracle and then see the customer deploy SAP on the Oracle platform. Oracle and SAP are always likely to optimize their applications for their own proprietary middleware, since their agenda is to expand their footprint inside large corporations, yet by doing so they rule themselves out of being an idealapplicationsn-neutral layer that can genuinely help their customers.

For most enterprises, Netweaver looks superficially attractive but does not solve the core integration problem, that of resolving the differences in business meaning embedded in their many, many core transaction systems. Netweaver's widespread deployment is certain, but its skin-deep level of integration will deliver only limited benefits to customers, yet at considerable cost. Perhaps customers should check back on the investment cases they made a few years ago before they went down the ERP route - did the benefits promised then actually materialize? The costs certainly did (billions of dollars each for global corporations), but I haven't seen too many of their IT departments shrinking away to nothing because all their integration problems were solved.


Post a Comment

<< Home