Insights

Q&A with the Microsoft Configuration Manager Product Team: Part 2 of 2

by

MMS 2016 Configuration Manager Q&A

At last month’s MMS conference in Minnesota, I got to learn a lot about System Center Configuration Manager (SCCM): past, present, and future. One very special session was the Q&A with the SCCM product team. I published Part 1 of my writeup of that session a while back, and this is Part 2.

Microsoft Configuration Manager Product Team members presenting included: Aaron Czechowski, Senior Program Manager; Kerim Hanif, Senior Program Manager; Dune Desormeaux, Program Manager; Kerwin Medina, Senior Software Engineer; David James, Principal Director of Software Engineering.

Answers are summarized, not a word-for-word transcript.

Question:

Do you support the use of task sequences for deploying applications?

Answer:

It is supported. However, app models really should be the tool for apps. If you are pounding a screw in with a hammer, then there is something to fix. If this is the best way to meet you need right now, then do what’s needed and give feedback to the dev team so we can give you better tools in the future.

Question:

What is the road map of ConfigMgr to the cloud? Exchange went online, SharePoint went online, what about Configuration Manager?

Answer:

Brad Anderson talks about this a lot. Sophisticated management (SCCM) is going to be around for a long time. Meaning at least 10 years. Some of that functionality will be available in the cloud via InTune, but the cloud is a nebulous thing. The mission for Microsoft is to give customers the tools they need to solve their problems, and the flexibility to choose what tools they want. Right now you can do mobile management three ways: SCCM + MDM on premise, hybrid SCCM + InTune, or fully cloud via InTune. That is a choice that will continue to be there. There is not an effort that says, “move everyone to cloud.” The two products, SCCM and InTune, will work together more and more often.

Question:

PKI and Pull Distribution Points. Can you guys make it so I don’t have to run a script to turn of PKI for HTTPS?

Answer:

You should put this on user voice. User voice is a very powerful tool and we use it very regularly. Use the comments as well to discuss in more detail. We read the comments to try to understand what people want specifically.

Question:

Do you have a timeline for supporting delta Windows updates?

Answer:

This is a big priority from the Windows team, and will be the model for Windows 7 and 8 as well, so there is a lot of pressure. But there is no timeline that I can commit to today.

Question:

Is there any thought about moving from DPs from one site to another, particularly useful in cases where a site server has died. We would like to take a DP, move it to a new site, and have its content intelligently mapped to the new site. Then we might want to move it back later.

Answer:

If the need is arising primarily in scenarios where a server has died, then my inclination is to try to understand, “Why do have a scenario where a server is offline?” When I look at high availability, that is the more pressing question in my mind—preventing the failure in the first place.

Question:

What is your culture like today? How are you operating? What is your mentality from an engineering perspective?

Answer:

About 18 months ago we took the combined SCCM/InTune team and separated the SCCM team from the InTune team. We thought about sophisticated management (SCCM). When we looked at Windows 10 coming down, we realized that in order for Windows 10 succeed in enterprises, SCCM has to facilitate that success. There will be a lot of versions coming, between the modern world of more devices and much more frequent updates. So, we had to rethink it. It’s a business continuity issue. Your business depends on the ability to patch and deploy. The key to that model is things like user voice and telemetry in order to keep pace with what customers need and what they prioritize. The goal is to keep up with a moving world.

The upgrade of 1602 was successful for most people who have done it. Our model is, “go fast without compromising quality.” The challenge is features that have been out there for a while, like patch and app deployment and OSD, these need to be extremely reliable features, version 6 or version 7 features. When we add new features, they’ll show up as version 1 features, they won’t be the same quality as features we’ve been shipping for a decade. For example, in order to build new features like Windows Store for Business, we can’t lower the quality of patch. For conditional access, we use the previews. We keep the old features at high-level quality, and improve the new features rapidly with the preview model.

From SMS 2003 to SCCM 2007, that was four years for a new version. Now we have a new version every month!

Question:

Pull distribution points. Can you manage bandwidth with pull similar to the way you do regular distribution points?

Answer:

BITS is an outdated technology created in 1992. It’s not great at managing bandwidth, but can help in a lot of cases. However, a separate problem with your scenario is that throttling is very difficult when there’s a machine between the source and destination. Once you have more than one machine in the scenario, you can’t throttle the end-to-end connection, it’s terrible.

Question:

It would really be nice if scheduling was based on the time zone of each individual node, is that being looked at?

Answer:

I think about the scheduling all the time. I’m always thinking through models. SCCM can be improved here. We have all these different places where we set it. Sometimes we randomize it, sometimes we don’t. Sometimes we let you control the randomization, sometimes we don’t. And then we have patch Tuesday, which is the second Tuesday, but maybe what you need is the Friday after the second Tuesday. Time should be simple and deterministic. You should be able to say what you mean and have the system do what you mean. Right now, you have to know the feature: how it’s going to interpret time, whether it’s going to do it locally or not, whether it’s going to randomize it or not, and if it’s going to give you a window around it or not.

There is room for improvement in consistency, but it’s also complicated because different customers have different preferences.

Question:

You were talking about keeping quality high, and how that would affect changes made to some of the other features. Does telemetry play into some of your decisions about which features to work on?

Answer:

Definitely. We have 54,000 test cases. We have a test team that all they do is loop through though all those teste cases every two weeks in parallel with development. With all the new features, we ask, “How do we bring the companies forward without disrupting them?” So, yes, we use all the information available to us, including user voice and telemetry, and we test constantly and very thoroughly.

Bill Bernat
Director of Product and Content Marketing, Adaptiva