Running a Service Delivery Organization…..

Hi folks,

I am Arun Chengi and I manage a team of very talented individuals who work tirelessly in delivering various software delivery services at Microsoft and to several external customers we manage.  I have been in the IT industry for nearly 20 years and been in Microsoft over 6+ years. The organization I am in - Management Platforms and Service Delivery (MPSD) - is a sub-group under MSD – the team responsible for developing System Center products and services. We use Configuration Manager as our primary delivery mechanism in managing a large client base of about 280,000 PCs at Microsoft. At this moment, the load is equally split between our 2007 and 2012 hierarchies, but in November we shall be an out and out Configuration Manager 2012 shop. In addition, we use just about every other product in the System Center stack, such as System Center Orchestrator and other MDOP products such as MBAM, App-V, , etc.

Our team’s charter includes delivering services to a large client base within Microsoft.  While we operate similar to any other IT shop in terms of meeting necessary delivery commitments in the form of patch, anti-malware, and OS and software deployments, a big part of our organization’s challenge/ contribution comes in early adoption of Microsoft products through what we very aptly term as DOG FOODING.  It is almost self-explanatory that if we don’t eat our own dog food initially, who would? 🙂

Seriously though, we take lot of pride in this work. Through this early adoption of technology on a large scale (which varies depending on the particular milestones of the product) we strive to provide feedback on the product quality directly to the product group and indirectly through end users who use these products.

A common question I get asked by my friends outside of Microsoft, as well as customers we interact with is how we work to manage and secure clients. Well, the simplest answer is likely close to how you all do it! But the real fun (hmm…for the most part) is doing the production deployments with products that are in early beta releases and still meeting the required compliance SLAs and goals that we signed up for. 

Let me be frank and admit that it can be challenging at times. Maintaining the service commitments and dogfooding at the same time is a tightrope walk. This is made even more challenging when we dogfood the very core platform that is responsible for delivering these programs: Configuration Manager. Even though offering early releases of Windows, Office, etc. has its own challenges, they are mostly user driven. What I mean by this, is that to a large extent users can opt-in or out of such deployments based on how risk averse they are. 

But when it comes to Configuration Manager, it is a different ball game altogether. While we do take deployments of this nature through incremental steps, by the Release Candidate stage we will in all likelihood be all in. We take every release through the necessary quality gates to make sure that in its early stages the product lives up to the quality that is capable of adhering to shared goals we develop with the Product team and our business stakeholders. My team has been spending a lot of time driving feedback to our product team by looking at every single feature of this product. We have been working very hard in validating Configuration Manager 2012 since its infancy. Just to provide some perspective, here is how we scaled through some of the major milestones of the product:

Beta1 Escrow – 10,000 systems

Beta1 – 50,000 systems

Beta 2 Escrow – 150,000 systems

At Release Candidate, we will deploy to our entire managed base – around 280k systems 

We are quite excited about this product. Configuration Manager 2012 has a great deal of new features. My favorite is its shift towards a user-centric design. The Software Center and Application Catalog enables users to choose and set certain criteria and options on how their systems should be ‘managed’ if IT chooses to honor it. Yes, a focus on IT control or IT governance is an option that an IT manager can now exercise. And there is an application model that now enjoys the same kind of features that patch management has had to a large extent, along with rich sets of detection options to deploy applications. This is expected to not only help remove custom tools (and we all have them, don’t we?), but also to help reduce the TCO and maintain SDLC. 

There are some other great additions - including fundamental changes in the replication model that improve upon the core of the system.  Also, if you are as passionate as I am about “Client Health”, you are going to enjoy the new agent called CCMEVAL that monitors the health of the client and other associated dependencies.  

A key thing to note here is that we do take user experience very seriously. In many situations, users in Microsoft are much harder on us than at several of the companies I have worked for in the past. This is very fair for the nature of business we are in. No user welcomes a random reboot or unnecessary popup, no matter what role they play in an organization. It is not any different in Microsoft but the users are often tougher – after all, at Microsoft everyone thinks they are a developer! On a lighter note, have you ever heard a user say, “Hey my machine was rebooted by a patch but my productivity wasn’t lost and I really appreciate IT’s intent in keeping my machine secure.” No?  Me either…I came from a company where unless proven innocent, it is always the SCCM guys creating the issue!!! Ok, I am being bit melodramatic here, but as I’m sure you know, it is not too far away from reality. 

The point I am trying to make here is that every agile organization needs to keep thinking about the end-user, and should always be on its toes in managing a vast client base. For an organization such as ours - which has an unequivocal passion towards protecting user experience and at the same time donning the hat of security police – it is a challenge.  I always think user experience is more an art than science. As the 'Consumerization of IT' picks up, user experience is going to continue to increase in importance and become an equal partner to SLA.  In fact, maintaining SLA and providing a great user experience will be equally important to IT Admins.  We should make sure that we are driving both of these pieces forward. All these years, enterprise management (and to some extent the product’s primary focus) was on the IT Admin - now we have 2 sets of personas to keep in mind (IT Admin & User).  Keeping a high bar for success is going to be really, really important as we move forward. And this is the journey upon which my team and I continuously strive to find that magic wand which says “Hey Arun, here is your formula”. 

That’s it for this ice-breaker post. In upcoming posts I will blog more about our portfolio of services and how we do IT from both an engineering and operations perspective.  You should also expect to see more SMEs from my team posting deep dive talks on some of these services that they support. 

I look forward to your feedback and your areas of interest in this space so that we can cater to those on a priority basis.

Comments (2)

  1. Arun Chengi says:

    I’m glad you enjoyed the blog post.  Let me know if you’ve got other Configuration Manager admin related discussion topics you’d be interested in or for my team to discuss to help you prepare better. You can go out to Connect to learn more about System Center Configuration Manager 2012.

  2. HeyAdmin says:

    Great piece Arun.  I'm on the brink of a re-design of our SCCM 2007 infrastructure.  If possible, I'd like to wait for SCCM 2012 RTM, but can't wait too long.  Any projections on the date of its release?

Skip to main content