none
SCCM Design RRS feed

  • Question

  • Hi all,

    I have an odd setup where I am in charge of an estate covering a whole country.  I have nearly 1000 remote sites - each of which only has upto 50 devices stationed onsite.

    For the sake of 50 devices... at a site, only 5-10 of which will be reimages per year - it seems pointless placing a DP or Secondary Site at each location given that the sites are only provisioned with a dodgy broadband line each.  That said, if I were to leave a single or pair of site servers at head office - I would be reimaging 5000-10000 devices per year over slow broadband lines as well as trying to look after update pushing, app pushing, hardware auditing, remote controlling machines etc.

    Suggestions on a good design would be appreciated!  So far, it has been sugested that I create a hyper-v DP at each site - but I was hoping to avoid 1000 DPs all trying to sync.  Confused!

    Tuesday, February 25, 2014 4:13 PM

Answers

All replies

  • Could you combine some places to use the same DP? You'd have to check your network configuration (which is linked to which remote site, or are they..)... 

    Tuesday, February 25, 2014 4:16 PM
  • Could you pls give us the exact location wise device count?

    Take this as a reference: (Considering Good Network Bandwidth)

    A regional site having: 40-250 clients

    Place a DP. If OSD Inscope , Place PXE enabled DP

    A regional site having: >=250-3000 clients

    Place a Secondary Site. If OSD Inscope , Make sure that you have PXE enabled DP here


    Thanks, Prabha G


    • Edited by Prabha G Tuesday, February 25, 2014 4:21 PM
    Tuesday, February 25, 2014 4:19 PM
  • You might also have a look at 3rd party tools (1E Nomad or Adaptive Onesite).

    Torsten Meringer | http://www.mssccmfaq.de

    Tuesday, February 25, 2014 6:03 PM
  • For the sake of 50 devices... at a site, only 5-10 of which will be reimages per year - it seems pointless placing a DP or Secondary Site at each location given that the sites are only provisioned with a dodgy broadband line each.

    First off, with the release of 2012 it has been rendered almost redundant to put Secondary Sites out on your network as it adds complexity that doesn't need to be there if it doesn't have to be. Secondly, if you have dodgy connections the last thing you want to do is have everyone coming back to head office for everything. The great benefit of DPs is the localization you get. You can also schedule data transfer windows so you can have data pushed to your DPs overnight while the offices are dormant. This way they don't saturate your already bad connections (trust me you aren't the first one to deal with this :-) )

    That said, if I were to leave a single or pair of site servers at head office - I would be reimaging 5000-10000 devices per year over slow broadband lines as well as trying to look after update pushing, app pushing, hardware auditing, remote controlling machines etc.

    This, I do believe, would give your network team an absolute nightmare. Not to mention that DPs only allow so many connections at a time to sort of "queue" requests as they happen. If you don't present enough DPs to your organization these 10k machines could be in a MASSIVE line to receive this stuff. Think Drivers License Office on free License Giveaway Day.

    So far, it has been sugested that I create a hyper-v DP at each site - but I was hoping to avoid 1000 DPs all trying to sync.  Confused!

    Creative ideas like this are never a bad idea. Especially where you can utilize underutilized resources. I remember I was at a company where we used CISCO WAAS devices to cache local data requests for slow networks. One of the benefits was we were able to host VMs on the WAAS to present AD services locally as well. Well, I got with the network team and they were happy for us to host an SCCM DP there as well. The DPs don't do much so they don't need much in hardware, just a lot of storage space and good throughput. As for the 1000 DPs trying to sync, you can use Pull Distribution points. Which basically allow you to trickle down distribution. You take away the "all roads lead to rome" approach and start segmenting it out. You could make it something like "Head Office is Manchester. Big cities like London could pull in from Manchester. Then smaller suburbs like Camden, Croyden, etc... could pull in from London." Thereby alleviating some network traffic. Keep in mind this would probably be governed more by your network than a map but most of the time that tends to follow the same pattern.

    Joyce is right as well. BranchCache is always an option. It allows for your clients to be your local DPs. But I personally like a dedicated resource I can count on and monitor. I think BranchCache can be utilized but it's hard to monitor uptime and count on endpoint devices that users control. That's just me though.

    In my opinion, if I were you. I wouldn't be afraid of the localized DPs as they will ultimately save you time and bandwidth on deployments. And just remember, the more a company uses this product the more the need for it grows. So it doesn't hurt to poise yourself for growth now rather than struggle for it later.


    Dustin Estes - MCP | www.dustinestes.com

    Wednesday, February 26, 2014 3:29 AM
  • Hello Ben, I hope you found an answer to your question. Don't forget to mark an answer that best resolved your question.

    Dustin Estes - MCP | www.dustinestes.com

    Thursday, February 27, 2014 12:37 PM
  • You might also have a look at 3rd party tools (1E Nomad or Adaptive Onesite).

    Torsten Meringer | http://www.mssccmfaq.de

    I agree. This could be the best solution for you. I have a customer with hundreds of small sites (10 - 20 PCs). They manage the entire environment with a single ConfigMgr 2012 management point and distribution point - integrated with 1E Nomad Branch, PXE Lite and NightWatchman.

    http://www.1e.com/it-efficiency/software/nomad-enterprise-software-deployment/



    Gerry Hampson | Blog: www.gerryhampsoncm.blogspot.ie | LinkedIn: Gerry Hampson | Twitter: @gerryhampson

    Thursday, February 27, 2014 2:21 PM
  • I've just done a blog series, "ConfigMgr 2012 - Integrate with 1E Nomad". It may be helpful for you.

    http://www.gerryhampsoncm.blogspot.ie/2014/03/configmgr-2012-integrate-with-1e-nomad.html



    Gerry Hampson | Blog: www.gerryhampsoncm.blogspot.ie | LinkedIn: Gerry Hampson | Twitter: @gerryhampson


    Monday, March 3, 2014 11:12 AM
  • What is considered "good network bandwidth"?  Is there a number where it makes sense to come back to the central location for the DP?  We have a lot of Gig connected sites, some with small client counts, some large, and we are trying to come up with criteria on when to place a DP, when to use Branch Cache, and when to just allow the traffic to go across the WAN.  As another person mentioned below, we are not even considering secondary sites with SCCM 2012.
    Wednesday, April 9, 2014 6:33 PM