My blog has moved and can now be found at http://blog.aniljohn.com

No action is needed on your part if you are already subscribed to this blog via e-mail or its syndication feed.

Sunday, January 7, 2007
« Attend M.I.T. for free! | Main | Relationship between Enterprise Architec... »

I was reading Todd's entry on "Services for Managing the Network" in which he comments on an article by the F5 folks which talks about a unified way to manage both services and network components through web service interfaces. 

This is something that I've been thinking about for a while as well.  To me a SOA runtime infrastructure should allow us to monitor, manage and administer across network, computing and service resources using standardized policies. In this ideal world, the appropriate domain experts (Security, Networking, QoS, SLA etc.) define the policies for that domain in a centralized manner, push out those polices to distributed appliances and service platforms across your Enterprise such that they can be enforced, and provides the ability to collect metrics on what is going on in your environment.

In some ways the greater challenge is not technical, but cultural. It lies in trying to provide a common frame of reference and understanding to folks who come from different background (NetOps/Transport folks, DataCenter/Computing Infrastructure folks, Service folks) on the impact of deploying a SOA runtime infrastructure. It is a challenge that I face on a regular basis and one that requires the most fundamental of skills - Communications and the ability to see the other person's point of view.

On the technical side of the house, the challenge is wrapped up in the phrase "standardized policies". To reach this stage requires two separate things to happen:

  1. The ratification of standards that address these various aspects of management
  2. Adoption of these standards by various vendors at the Network, Computing and Service layers

At the current stage of technology, some of what I am discussing above is possible by using a combination of WSM, Mediation Systems, Registry/Repository, Platforms and Network Devices. But unfortunately, given that a lot of the standards are not finalized, it requires one to use specific vendor products (where vendors have established interop relationships) that are using proprietary mechanisms in the absence of established standards.

So what are some of the standards that we should be tracking and urging our vendors to support in the policy and management space?

  • Policy - WS-Policy to start with. But keep in mind that WS-Policy is simply a container and still requires the creation of multiple domain specific languages that will address areas such as SLAs and QoS etc.
  • Provisioning - Adoption of Service Provisioning Markup Language (SPML) v 2.0. Keep in mind that this deals purely with user provisioning and not with service provisioning. Current service provisioning is, to a great extent, a manual process.
  • Management & Reporting - The convergence of WSDM and WS-Management. Note that this has a dependency on the convergence of WS-Eventing (WS-Management needs this) and WS-Notification (WSDM needs this) into WS-EventNotification.

As you can see above, there is a lot of work that still needs to be done in this space and significant competition among the various vendor factions regarding what these standards should be. As you are building out your infrastructure, I would highly recommend that you question your vendors on their support for existing standards, their tracking and participation in the standards process, and their roadmap for support of future standards, so that in the end you have the ability to monitor, manage and administer your environment in a holistic manner.

1/7/2007 1:47 PM Eastern Standard Time  |  Comments [0]  |  Disclaimer  |  Permalink   
Comments are closed.