June 2006 Archives

SOA and SLA

| No Comments | No TrackBacks

As architects and developers pull together more and more services into larger services and business processes a much closer examination of the quality of those services and the service level agreements (SLA) must be made and documented. The throughput and uptime, the types of bindings and transports, the capability of participatingin a transaction or providing a certain level of security are all examples of service level agreements that extend beyond just the schema contract.

Consider a larger business process that is made up of a half a dozen services. If any one of the services as part of the business process fails to meet certain criteria then the entire business process might fall below it's SLA. A service that claims a certain SLA today must continue to provide the same level of service tomorrow in order not to effect the service level of the composite services and business processes it is a part of.

This is where all the governance that people are talking about comes in. For early SOA adoption this type of governance might not make sense but as the extent of your SOA grows and services start to be reused more and more then governance is required and great care must be taken to examine each services SLA and provide enough information so that other services can consume it with confidence.

So is all this some tedious effort to just come up with more documentation? More standards? More XML? On the contrary SLA and governance becomes a powerful ally in analyzing, developing and testing new services. So much of the cost of making changes to your business are about the effects on existing assets. Understand the SLAs across the various services provided by IT helps testing services as part of composites and business processes and, more powerfully, help reduce the cost of regression testing in future development life cycles. After all that is where much of the business agility promised by SOA comes in - the ability to adapt your IT to changing business needs at greater speed and efficiency.

There are a number of initiatives that are trying to address governance and SLA. Notably the W3C's WS-Policy.

I wrote a blog entry back in January about using RSS as a cheap way to do a services registry. I'm told by one source that as a result several hundred downloads of Celtix were generated in the first week of that posting. Since then this posting is still the most read entry on my blog. The second most well read posting on IPBabble.com is my posting on JBI.

I've received a lot of positive feedback on this article. Many asked me why I didn't come out stronger against UDDI. I've noted since then that there seems to be a lot of activity around using RSS for a registry. I think it can be used as is today for small numbers of services but some enhancements are required in order for it to scale. More specifically: standardization, tooling, and how to allow it to scale (federation).

But it does raise the questions: Why is UDDI not just killing off this sort of ad hoc approach? Where is UDDI in terms of adoption? Are people buying it? One would think that it would dominate the market around services repository but it hasn't. It may dominate the commercial repositary space but there are a lot of ad hoc approaches going on in SOA implementations.

Some industry leaders I've spoken too simply don't like UDDI. "It's too hierarchical", "It's too much like a CORBA Naming Service", "I need something that's more flexible in terms of lookup". For many, as I mentioned in that earlier posting, it's just too big for the few services that they have started with. Others seem to be aligning with my view of using technologies like RSS to provide a cheap and easy to use registry. (And if the hits on my original post are anything to go by we should see much more.)

In my original post I tried to stay neutral on UDDI saying:

"I am not trying to replace UDDI. UDDI is the right and standard approach for discovering Web services. In fact I think that my idea can compliment UDDI ..."

Though I still maintain that the RSS approach can compliment UDDI, I'm leaning more towards the "who needs it" (UDDI) approach. This is based on the response I've received. And I'll need stronger convincing that UDDI is necessary. Perhaps UDDI needs an overhaul?

Who is IPBabble

William Henry IP Babble is the personal blog of William Henry.

William has 20 years experience in software development and distributed computing and holds a M.Sc from Dublin City University. He is currently working in the office of CTO at Red Hat on the MRG product. This weblog is not funded by Red Hat.

Posts are intended to express independent points of view, but understand that there is probably a bias based on the influence of working with standards based middleware for over a decade. (See disclaimer below)

February 2009

Sun Mon Tue Wed Thu Fri Sat
1 2 3 4 5 6 7
8 9 10 11 12 13 14
15 16 17 18 19 20 21
22 23 24 25 26 27 28

Disclaimer

The views expressed in this blog are solely the personal views of the author and DO NOT represent the views of his employer or any third party.

About this Archive

This page is an archive of entries from June 2006 listed from newest to oldest.

April 2006 is the previous archive.

July 2006 is the next archive.

Find recent content on the main index or look in the archives to find all content.