3 Replies Latest reply on Nov 29, 2012 1:51 PM by TomHowarth

    Fault tolerance shared files

    engyaz Novice

      hi all.

      i am confused about fault tolerance. i am wondering if you can answer my questions point by point

       

      if i enable fault tolerance for A VM i will have a secondary VM in another host.

      my questions are:

       

      1) the files of the 2 VMs (primary and secondary) are seperated so they are not sharing the same files?  in other words if the files of the primary VM are corrupted the second VM can take the load?

       

      2)what if the operating system on the primary VM is corrupted the secondary virtual machine will not boot also?

         for example

          when i  start the primary VM i will get for example windows NDR missing and  black screen. the secondary VM operating system will stay running or

          i will get the same situation as on the primary?

       

      3) in fault tolerance everything running on the primary virtual  machine is running also on the secondary virtual machine with a small delay

         so if the primary VM get a virus the virus will run also on the secondary VM?

       

       

       

      4) i read in some document that the 2 VM share the same files. if so

      why when i create the secondary VM more space are taken on the DATA Store?

       

      5) if they share files what is shared and what is not shared?

       

      5)if what i said before about if an error occupred on the primary VM will occure also on the secondary one is true so where is the importance of fault tolerance? i am not getting in what it can protect the VM

       

       

      thanks in advance because i am not finding answers to my questions

        • 1. Re: Fault tolerance shared files
          vmroyale Guru
          User Moderators

          Note: Discussion successfully moved from VMware ESXi 5 to Availability: HA & FT

           

          1) the files of the 2 VMs (primary and secondary) are seperated so  they are not sharing the same files?  in other words if the files of the  primary VM are corrupted the second VM can take the load?

          They are using the same files on shared storage - this is a requirement to use FT.

           

          2)what if the operating system on the primary VM is corrupted the secondary virtual machine will not boot also?

             for example

               when i  start the primary VM i will get for example windows NDR missing  and  black screen. the secondary VM operating system will stay running  or

              i will get the same situation as on the primary?

          The machines are in vLockStep and using the same disks on shared storage, so anything that happens to one will happen to the other. A BSOD or other failure will occur on both VMs simultaneously.

           

          3)  in fault tolerance everything running on the primary virtual  machine  is running also on the secondary virtual machine with a small delay

             so if the primary VM get a virus the virus will run also on the secondary VM?

            Yes it will.

           

          4) i read in some document that the 2 VM share the same files. if so

          why when i create the secondary VM more space are taken on the DATA Store?

          There are files that will get created when the secondary VM is powered on - like the SWAP file, logs, etc.

           

          5) if they share files what is shared and what is not shared?

          The virtual disks that the VM uses are what is shared between the two. SWAP files, logs or other VM-specific files are unique, but the OS doesn't know about these files.

           

          5)if  what i said before about if an error occupred on the primary VM will  occure also on the secondary one is true so where is the importance of  fault tolerance? i am not getting in what it can protect the VM

          FT protects your applications from ESXi host failures. Where HA could restart the VM, FT can "resume" it with the current state on another host.

          • 2. Re: Fault tolerance shared files
            EdWilts Expert

            The traditional use of FT, when it was all done in hardware, was to not use FT for everything because it's simply not needed.  Even banks didn't use FT on their database servers.

             

            You could use FT on a front-end server that the users connect to but isn't application-heavy.  The light-weight app and would be designed to never fail and the OS would be hardened so that virii and worms would not get in.  Transactions would be held in the FT farm before being committed.   Behind all of that would be your beefier application and database farms.  An individual node could fail and you wouldn't lose a transaction - it would be retried on another host in the farm.  After all, you don't want your bank to lose your deposit, do you.  The back-end servers also commit the transactions to multiple disperse storage pools

             

            FT has a HUGE overhead.  Most applications simply don't need it because that's simpy not where they fail.  As you've pointed out, applications will crash and FT does NOTHING for them.  If you do a Windows Update on an FT guest, you'll apply the patches to both guests at the same time, and a reboot will. reboot both of them.  That's normally not what you want.

             

            You need to determine where FT is right for you, where HA and DRS rules are right for you, and where you simply need farms of back-end servers that can take the rolling reboots without taking the applications offline.   This isn't trivial.

            • 3. Re: Fault tolerance shared files
              TomHowarth Guru
              User ModeratorsvExpert

              engyaz wrote:

               

              hi all.

              i am confused about fault tolerance. i am wondering if you can answer my questions point by point

               

              if i enable fault tolerance for A VM i will have a secondary VM in another host.

              my questions are:

               

              1) the files of the 2 VMs (primary and secondary) are seperated so they are not sharing the same files?  in other words if the files of the primary VM are corrupted the second VM can take the load?

               

              The Primary and secondary guests share the same VMDKs on shared storage,  if a coruption occurs on the primay machine resulting in a Blue screen or actual data coruption then it will occur on the secondary machine too.

               

              Fault Tolerance is not a DR solution but a High availability, if you need data protections then look to another solution.

               

              2)what if the operating system on the primary VM is corrupted the secondary virtual machine will not boot also?

                 for example. 

               

              when i  start the primary VM i will get for example windows NDR missing and  black screen. the secondary VM operating system will stay running or i will get the same situation as on the primary?

               

              If you are starting from Boot the machines will not start and as they are in lockstep and sharing the same VMDK'st there will be no failover

               

              3) in fault tolerance everything running on the primary virtual  machine is running also on the secondary virtual machine with a small delay

                 so if the primary VM get a virus the virus will run also on the secondary VM?

               

              Yes because they are in lock step.  whatever instructions are processed at the primary machine are transfered to the secondary machine.

               

              4) i read in some document that the 2 VM share the same files. if so

              why when i create the secondary VM more space are taken on the DATA Store?

               

              there is extra space taken as the are two sets of VMSS files for Memory management, but both machines share the same VMDK;s

               

              5) if they share files what is shared and what is not shared?

               

              VMDK, VMX, but not VMSS

               

              5)if what i said before about if an error occupred on the primary VM will occure also on the secondary one is true so where is the importance of fault tolerance? i am not getting in what it can protect the VM

               

              You are protecting the Guest from the failure of a VM Host. this technolgy is used where you cannot suffer an outage of the time it takes for HA to restart a guest on another host after a host failure. it is not Data Protection but Uptime protection.

               

               

              thanks in advance because i am not finding answers to my questions