3 Replies Latest reply on Aug 30, 2012 3:11 PM by dsbalcau

    Agent requirements for storage monitoring and chargeback/showback

    amozen

      I am new to the product and want to go for try out. So my question is: do I need to install agents on all SAN attached servers to be able to do chargeback/showback as well storage monitoring or just install several VMs, deploy agents on them and collect data from my storage using SMI-S. I am a little bit hesitant to deploy agents on all SAN attached servers, especially as all of them running EMC Control Center agents.

       

      Thanks fr advise.

        • Re: Agent requirements for storage monitoring and chargeback/showback
          dsbalcau

          Hi amozen,

           

          No, you should not need to deploy agents to all of your SAN attached servers. The host on which you install the main Storage Manager server should be able to communicate with your servers running the vendor's SMI-S provider. As you significantly scale out your system, it is advisable to install agents for load balancing, but you haven't given me any indication of the size of your system in your post.

            • Re: Agent requirements for storage monitoring and chargeback/showback
              amozen

              Hi Balki 1429,

               

              The size of the environment is quite big. We have about 20000 disks in total (40 frames across 9 data centers in different continents). I've read STMAdministratorGuide and search forum, but cannot find answers to all my questions.

              Is it possible to set up Storage Manager is a way where it's services distribute among several servers?

              Does Storage Manage supports load balancing?

              Can one agent, you suggested, pull the data from different locations? What is the maximum network latency supported between the agnet and SMI-S provider?

              What is feasible server size in regards of memory / and CPU count for such environment I need to plan? The document mentioned 8 to 16 CPU for big deployment, are those actual CPUs or Cores?

              And last but not least question. If I want to see host file systems utilization I do need to deploy agents on SAN attached server. Am I right? Does your agent interfere in any way with EMC Contol center agents, we have installed on every single host in  our environment?

               

              Thanks,

              Alex

                • Re: Agent requirements for storage monitoring and chargeback/showback
                  dsbalcau

                  HI Alex,


                  Your questions were rather detailed, so I've responded inline. Hope this helps!


                  The size of the environment is quite big. We have about 20000 disks in total (40 frames across 9 data centers in different continents). I've read STMAdministratorGuide and search forum, but cannot find answers to all my questions.

                  Is it possible to set up Storage Manager is a way where it's services distribute among several servers?


                  We can distribute the collection from the arrays across multiple agents, but services of the Storage Manager server cannot be distributed.  We have some customers that have moved the mysql instance to another server, but this scenario is not handled automatically by upgrades.


                  Does Storage Manage supports load balancing?


                  No, Storage Manager has no built-load balancing.  You can distribute the collection load across multiple agents.


                  Can one agent, you suggested, pull the data from different locations? What is the maximum network latency supported between the agnet and SMI-S provider?


                  Yes, one agent can pull data from different locations assuming the network connectivity is available, but we suggest putting agents at each location to collect the data locally, which reduces traffic on your WAN.  When requesting the data from the provider, the agent will wait for a certain amount of time (generally in minutes) before timing out.  Network latency generally is not a problem.


                  What is feasible server size in regards of memory / and CPU count for such environment I need to plan? The document mentioned 8 to 16 CPU for big deployment, are those actual CPUs or Cores?


                  8-16 cores (not sockets) and more RAM is better, as is faster disks.


                  And last but not least question. If I want to see host file systems utilization I do need to deploy agents on SAN attached server. Am I right? Does your agent interfere in any way with EMC Contol center agents, we have installed on every single host in  our environment?


                  For VMware, we collect file system utilization via vCenter API.  The host machine on which Storage Manager Server is installed does not need a separate agent installed; however, for physical servers, you do need to deploy agents.  We have no known problems with EMC control center agents.