System p Entry Points for SOA

Well the wagon has wheels, one of the first visible results of the work I’ve been involved in System p was announced last week via press release.

The “System p Configurations for SOA Entry Points” are a collection of reference architectures, installation, system setup, configuration guides, as well as certification of the Software stack on System p, common integration patterns, best practices for problem prevention, role specific documentation, answers to common operational questions and appropriate customer-use cases. [BonusPak anyone?]

For me the benefit of a virtualised infrastructure to SOA and web services always seemed obvious and not just by virtualising at the middleware layer. Virtualising at the middleware layer is good for business applications and business services, it isolates them from the underlying technology and especially changes in that technology. Apps and service written in Java have the obvious advantage of being portable between systems and virtualized by the fact of executing in the Java Virtual Machine (JVM).

In some respects, and I had this discussion with Colin Renouf at lunch yesterday, it’s easy to assert that middleware like WebSphere is becoming the new operating system. It would be wrong though. There are hundreds and thousands of events, activities and actions that need to be taken to run computer software on computer systems. Not least interfacing to things like devices, storage and other I/O.

Forcing device drivers into middleware would be a mistake, it would make it unreliable and open it up potential security issues which just don’ exist today. Then there is the other 95% of things that operating systems do that either doesn’t belong in middleware, or would slow it down and make it less reliable.

So thats why virtualising the infrastructure outside of the middleware makes sense. System p already has some of the most robust virtualization technology, we have many customers using 50+ logical partitions on a single machine and sharing resources to run those partitions. This reduces energy requirements compared to stand-alone servers and also reduces the heat produced by running them, which in turn further reduces the power requirements.[See my Linux Consolidation entry]

So apart from  dealing directly with the hardware, and virtualising it, what other benefits can this bring. Well for a start, if your SOA middleware is running in a logical partition, or virtualized, you can virtualise the database or transaction server that your business services might be using as a backend. You can do it on the same system or another. Then using workload management you can schedule and prioritise work based on service level agreements. It is often the case that you can do this in the middleware. However, what the middleware won’t understand, and won’t have any control of, is the other work running in another partition, let alone being able to control it.

Also, the ability to use some of the other APV features and upcoming announcements and deliverables to manage long running public web servcies, will be a further level of virtualisation and abstraction. Imagine for instance being able to migrate a long running service instance from one system to another without shutting it down or causing a user outage.

Amongst the coverage of the System p SOA Entry points was this on Grid Today.When the first System p SOA Entry points ship I’ll post more details and some customer feedback. In the interim, as always I’d welcome feedback and comments. If you don’t want to comment here feel free to email me.

Finally thanks to Pamela Yeamans and John Langlois who’ve been the driving force behind actually getting the entry points this far!

6 Responses to “System p Entry Points for SOA”

  1. 1 John Langlois March 16, 2007 at 3:29 pm

    I really appreciate the mention on this page. And I feel guilty for violating many of my own project management principles. We’ve never been formally introduced! After reading your ABOUT page, I now know much more about you then you do about me. In this crazy world of virtualized teams, that’s not good.

    My website listed above may give you some insight into my being … my skill set is completely different then yours though complementary. I already see the potential synergy in your discussion about how APV features can migrate a long running service. You speak to the use of the technology to support the SOA environment. And that’s a different and more relevant angle than selling hardware on pure speeds-and-feeds. We need to sell solutions that satisfy customer needs.

    Buckle up for safety. We’ve just started an SOA wildfire that will be difficult for all of us keep under control. Can we keep up with the demands that this solution set will put on our time? And still maintain work/life balance? You’ve gotta run. I’ve gotta care for my wife, daughter and two mutts and my interests outside of IBM. If we find the proper balance, this year could be great for both us.

  2. 2 cathcam March 17, 2007 at 8:33 pm

    Thanks John, I’m sure in due course we will meet up. One of the best things about virtual teams is that you get to work with the best people, irrespective of location!

    Not sure about you negotiation skills though if it toook you six years to propose, I met my wife and 3-months later we were enaged. 26-years later I’m starting over after a great time. Enjoy your kids, you’ll be surprised how quick they grow!

  3. 3 Colin Renouf March 21, 2007 at 6:00 pm


    Good site! Very helpful!

    I think I should clarify “the WebSphere is the new OS” into different areas…. with it all being about perspective and your point of view…

    Almost all of the enterprise applications/packages that previously would have been written to an OS suite of application programming interfaces are now written to a JavaEE set of application programming interfaces, with the enhanced enterprise level services that go with it. I can’t think of a single enterprise package vendor that either hasn’t already moved from native OS interfaces or announced their move. Whilst JavaEE environment has many of the features of an OS; i.e. the JVM has to schedule threads, manage memory, etc it will never access devices directly. That is always the realm of the real OS, but isn’t that the way that AIX is going with virtualised devices mapped into partitions that use the Virtual I/O Server as the manager for the real devices! So, from the application perspective the Java and JavaEE combination is the OS running on the Java virtual, rather than real machine. From the hardware perspective the OS is as it has always been. So, like Eclipse, its all about perspectives and views!

    So, as we virtualise services and the underlying applications that are used to provide and compose them, the underlying OS gets pushed further and further from the view of those who are running and using the services themselves. Some of the virtualisation and isolation features that either have been or are about to be released then help to achieve the desire of location transparency in that I can move the virtual service from one box to another as the physical resource utilisation dictate.

    SOA itself is also about perspectives and views. What one requirement might see as coarse grained service may be perceived as a fine grained service ripe for composition by another requirement. This is often missed in the rush to adopt a SOA environment. Virtualisation technologies assist in adding the flexibility to change and compose services without massively changing the physical infrastructure.

    The current WebSphere Application Server 6.1 technology stack is built from components that can be extended and that themselves extend other components. These are all based on the OSGI/Eclipse/Equinox extensions/plugins/bundles and support concurrent use of multiple versions of particular extensions and extension points. Why is this useful? Well, the application server technology can be extended to tie into the OS virtualisation technologies and system technologies – and be used to get rid of the holes in the security and compliance model caused by having a virtual machine duplicating services offered by the OS in a different way. The web services feature packs for WAS 6.1, Process Server, and ESB are all just the start of the building block approach to extending the base functionality to provide more features for services, business processes, and workflows that fit with an enterprise service bus model. Build this on a flexible, mobile, virtualised platform rather than one tied to a location and a particular physical box and you have something that is truly flexible and finally allows the logical business architecture to be isolated from the physical infrastructure architecture that underpins it.

    See you soon!

    Cheers mate


  4. 4 cathcam March 22, 2007 at 5:06 pm

    It’s interesting as the changes in middleware are mirrored in changes in OS and vice versa. My views on Grid and the OS are well known. When we first got into Grids(2001) it was with a view of adopting the “distrubuted OS” structure. So as you describe, we wanted to adopt the idea of a set of services to provide the functions of the OS.

    Thats where we are heading from an infrastructure management perspective, see my entries on WSDM for some more detail:

    We are not committed to delivering on this yet, but this is the direction we’d like to head. I also think that you’ll see the OS become more modular building block, service based, especially when you look at running hundreds of OS images in partition on the same machine. We can make the OS much more flexible by having it services based, with modular services available for all the OS images within a system and possibly across systems depending on latency etc. Especially when you add live partition migration into the mix as a method of delivering continously available services.

  5. 5 sandra742 September 9, 2009 at 8:34 am

    Hi! I was surfing and found your blog post… nice! I love your blog. 🙂 Cheers! Sandra. R.

  6. 6 angelina jolie September 10, 2009 at 10:08 am

    I love your site. 🙂 Love design!!! I just came across your blog and wanted to say that Ive really enjoyed browsing your blog posts. Sign: ndsam

Leave a Reply

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

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

Google photo

You are commenting using your Google 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 )

Connecting to %s

About & Contact

I'm Mark Cathcart, formally a Senior Distinguished Engineer, in Dells Software Group; before that Director of Systems Engineering in the Enterprise Solutions Group at Dell. Prior to that, I was IBM Distinguished Engineer and member of the IBM Academy of Technology. I am a Fellow of the British Computer Society ( I'm an information technology optimist.

I was a member of the Linux Foundation Core Infrastructure Initiative Steering committee. Read more about it here.

Subscribe to updates via rss:

Feed Icon

Enter your email address to subscribe to this blog and receive notifications of new posts by email.

Join 915 other followers

Blog Stats

  • 88,694 hits

%d bloggers like this: