4 Replies Latest reply on Jul 12, 2013 11:33 AM by joebaggadonuts

    OSD Slow?

    Rookie

      Hi all,

       

      We have run into an issue which we didn't come accross in the previous years. We have multiple Images on a SAN that we need to deploy to computer labs. Each lab has 25-20 computers. Nearly all have a gigabit link back to the switches. The image is NOT just the OS. It is a full blown image. The OS, all of the software needed and configured to our needs. This is how we've done it in the past as well.

       

      Typically, OSD would take about 2 hours, which was fine. All of a sudden, it's now taking near 7 hours (after upgrading to SP3)! Any thoughts?

       

      I know our guys are NOT using multi-cast but rather a direct download. The entire image after captured with LANDesk is 87GB. Any thought on what we can do to improve this time or make this process more efficient?

        • 1. Re: OSD Slow?
          MarXtar ITSMMVPGroup

          What network card drivers are you using inside WinPE? Have you updated them to Win7 drivers? I wonder if perhaps since WinPE has been updated you are no longer using optimised network card drivers.

           

          Mark McGinn

          MarXtar Ltd

          http://landesk.marxtar.co.uk

          LANDesk Silver ESP

           

          The One-Stop Shop for LANDesk Enhancements

          - Wake-On-WAN - Distributed Wake-On-LAN, Scheduled Power Down, and SWDist Sequencing

          - State Notifier - Real-Time Device & User State Inventory Updating & Alerting

          Update - WoW & State Notifier now integrate for even more functionality

          • 2. Re: OSD Slow?
            Rookie

            We are using whatever LANDesk pushes ot the Reps. We haven't tinkered with the PXE Reps other than deploying them through LANDesk.

            • 3. Re: OSD Slow?
              MarXtar ITSMMVPGroup

              If you have redeployed the reps then it will have changed from the Vista-based WinPE version to the Win7-based version. It may not be related, but try inserting the network card driver for your machines into the WinPE image, it may be it is more optimised than the standard one Win7 provides.

               

              Mark McGinn

              MarXtar Ltd

              http://landesk.marxtar.co.uk

              LANDesk Silver ESP

               

              The One-Stop Shop for LANDesk Enhancements

              - Wake-On-WAN - Distributed Wake-On-LAN, Scheduled Power Down, and SWDist Sequencing

              - State Notifier - Real-Time Device & User State Inventory Updating & Alerting

              Update - WoW & State Notifier now integrate for even more functionality

              • 4. Re: OSD Slow?
                Apprentice

                To JNick,

                 

                I have been dealing with some portion, if not all of the problem, which you have presented here. We were having a problem where the more units that we deploed during the provivioning process (the more computers that are in the task to deploy the OS and provision the machine(s)) the slower the image was laid down. We found that with one PC it took 12 minutes to lay down the 5 GB imge with Imagew v.2 and with 4 machines it took nearly 50 minutes. After a close examination we discovered that the port the target was plugged into was reporting a 10 MB FD connection. This was supported because no matter how many PCs were connected to the host and downloading that host nver showed more than 8 to 10 MB of data being sent out over its network connection.

                The problem (cause):

                   When systems PXE boot the network speed is set within their bios and this is usually 10 MB. Once the payload comes down and the computer boots to PE the system tries to load a NIC driver. Now there is a default NIC driver in the LANDesk PE build and that was what is connecting to our Catalyst switchs at 10 MB. The problem is the default NIC driver(s) in your PE.

                Our Solution:

                   We called our Lenovo support team and they gave us a list of drive3rs to try. They worked. When those drivers are added (and you have to load the correct one for EVERY NIC that you have in your fleet) then the catalyst reports 100 MB FD and the source preferred server now shows the image coming down at 85 - 93 percent of 100 MB. Now the time it takes to run ImageW v.2 did not speed up (there are things going on there that cannot be hurried), but now more than one system does not cripple the deployment. This allows us to simultaneously roll 8 systems per task and realistically provision some 200 machines per 8 hour period. I suppose if we hag a GB link we could really rock n roll, but 200 is nice.