4 Replies Latest reply on Apr 6, 2010 8:15 AM by SafeBoot

    endpoint encryption high availability \ Cluster

      Hi All,

       

      We just finished the installation of EndPoint Encryption for PC 5.2.4 with the EndPoint Encryption server manager and everything is working great.

       

      Managment would like to create a cluster or some sort of high availability if the server fails.

       

      Can Endpoint be setup in a cluster or High Availabilty ?

       

      If yes, can you please point me to the documantation or give me more info.

       

      Thanks again,

      Simon

        • 1. Re: endpoint encryption high availability \ Cluster

          All the docs you need are part of the Microsoft Cluster kit. There's nothing special required? You're thinking active-passive yes?

           

          That will probably mean you want to store your db on a SAN (sloooow). :-(

          • 2. Re: endpoint encryption high availability \ Cluster

            Hi,

            Thanks for the quick reply, Yes I was thinking about active passive.

            Can this be done without MS clustering ?

             

            Regards,

            Simon

            • 3. Re: endpoint encryption high availability \ Cluster

              What other clustering do you have in mind, if not MS? EEM database must be run on Windows.

              SAN might be sligtly slower than optimized DASD, but is seriously worth increased reliability and flexibility.

              You might be forced to use DASD if you plan to have more than 500,000 objects in the database.

              • 4. Re: endpoint encryption high availability \ Cluster

                There used to be a bunch of companies who made cluster software, and cluster file systems, and of course if you wanted to be obtuse you could run it under Wine, or virtualized on a linux cluster even. Active-Active would give you true HA of course, Active-Passive, limited benefits.

                 

                Not sure anyone has done it before in production - there seems little point to go so far. If the db goes offline for a few minutes, all the clients keep working - the worst case is you won't be able to do any password resets. High availability, ie microsecond downtimes seems excessive, though perhaps less so if you're using EEFF.

                 

                I built a SCSI cluster once, that was fast and robust, but I am alway suspicious of SAN based storage - I've never seen it perform anywhere near DAS. But, if you only have a few thousand nodes you're unlikely to notice the difference perhaps.