02/14/08 :: [SOA] SaaS: CRUD-Oriented Architecture? [permalink]
I stumbled on this report from GSX: SaaS, spurring innovation, enhancing business value and enabling business processes.
This is one of the better report I have seen on SaaS. As a starter, I would like to re-iterate my position on SaaS. SaaS is a transient state of the industry. It marks the beginning of an evolution (the utilization of the Web in the application model), it is not an end in itself. The problem with SaaS (as it is defined today) is that we are still using a CRUD-oriented monolithic architecture where business processes, resources and services are not explicit. Yes, from a business model, SaaS wins hands down compared to the hassle of implementing-deploying-operating-upgrading cycle. SaaS took all the costs incurred by software utilization and centralized them to drive economies of scale.
But,... SaaS gives you agility in one dimension only: upgradability. This is why I disagree with one of the belief of the report:
Software as a Service will be more profound than most industry observers currently assume.
SaaS is going to displace traditional business software models, no questions, but beyond? The forces that will crunch SaaS (as we know it today) are too big to be ignored.
Reporting across SaaS silos? How about BI-aaS? no, that won't work, why? the reason is very simple. What makes SaaS attractive is upgradability. This means that the SaaS provider is really in control of your information model. How are they going to be able to guarantee that any evolution of the system of record is forward compatible with your reporting infrastructure?
Integration across SaaS silos? IaaS? no that won't work either for the same reasons. You are going to trade off upgradability for stability. In other words, your ability to upgrade is bounded by the stability of integration with other solutions, not just by cost.
Last, but not least, business process innovation? You have now mostly surrendered your ability to innovate to the SaaS provider (remember, you just laid off your delivery team?).
The best proof of what I am saying is that one of the report's tag line is "Spurring Innovation". Sounds great to hear, but the report does not substantiate just once how SaaS is going to spur innovation, because the reality is that SaaS hinders competitive innovation. It might actually get you in a bigger mess than before because your SaaS provider is not going to provide you with SLAs around upgradability.
IMHO, SaaS is going to evolve rapidly in the next 3-5 years with the emergence of Business Process Platforms (IBM, Oracle, SAP), to become the system of record and the value add services that are assembled into business processes.
- Service versioning is far more practical as SaaS versioning
- Business activity monitoring and complex event processing is far more practical than BI in a SaaS world
- Business Process Platform do not require "integration", "integration" is part of the programming model
- Innovation is far more practical in a BPP / SOA world than in a SaaS world
Interestingly, the GXS report does not talk about "architecture" either (since SaaS is based on an old CRUD-oriented archtiecture), if they had spent about 30 seconds on this topic they might have come up to the same conclusions.