General discussion

  • Creator
  • #2131202

    Replication nightmare


    by redwine ·

    My company has around 75 W2K servers all of
    them are Domain Controllers. 4 of them are at the headquarters and the rest dispersed nation-wide. Trying to have replication working correctly it’s a full time job. Our C: partitions are 4 GB and arerunning out of space everyday. Stagging areas grow up to 500 MB and replication stops when one server can’t talk to another. We have 3 Admins dedicated to monitor all servers, even though we make it work I believe it is such a nightmare. My question is this the way is suppose to be or are we doing something wrong? Anyone has a similar experiece?
    All have SP2 installed. 2 DNS servers and 1 WINS server for the entire Enterprise.
    1 GB memory and 2 CPUS 900 MHZ. per server

All Comments

  • Author
    • #3569432

      Replication nightmare

      by azizhakim ·

      In reply to Replication nightmare

      From what I have learnt in my win2000 studies
      during my studies for my 70-240 test,here is my brief input.I am assuming you have a single win2000 domain.Infrastructure master fsmo should be placed on a non-GC server(inless all DC’s are GC servers)My thoughts is to put one GC server on each remote site.On remote sites have install wins servers for local resolution and point to your hq wins for secondary resolution(hub &spoke topology)Intersite Replication should be scheduled to offpeak hours,bydefault win2000 replication is set to always,or removed through scheduling when the DC’s are busiest with logons.View application logs where areas of problem can be isolated an attempt to resolve.In my case although my network is not a production enviroment but for studies,app logs have been very helpful
      in suggesting remedial actions,I’m sure you have checked it out & I hope I have been helpful

    • #3569323

      Replication nightmare

      by joseph moore ·

      In reply to Replication nightmare

      75 DCs is a lot. I would imagine you could trim this down some. Every Win2K server does not need to be a DC. Unless you are supporting hundreds of thousands of users/computers per office, then providing there is sufficient bandwidth between offices,you can start to carve things up.
      Set up Sites for the main locations of your organization. If there are fast connections between small offices and larger ones (T1, 128K ISDN), then you don’t need a DC in the small offices. Let the users in the small offices get their authentication over the line from a DC at the larger offices.
      If you need multiple DCs per office/Site, then setup the replication to go between Bridgehead Servers only. Designate 1 server at each remote location as a Bridgehead. This way, these servers are the ones that do the main replication. Then, the other DCs in each remote site will get their replication info from the Bridgehead. It cuts down dramatically on the cross-talk.
      As for drive space on your DCs, this is always an issue. REmove any junk on the C: drive you don’t need. Uninstall un-necessary programs. Delete BMP, WAV, TMP files. Move the swapfile from C: to another drive. Move the Temporary Internet Folders folder from C: to another drive.

Viewing 1 reply thread