The IBM Cisco VersaStack comes with a set of GUI's for the Cisco UCS and the IBM storage.
Both enable to manage the system but storage monitoring and analysis capabilities are poor.

This is like flying a plane without a cockpit!

BVQ adds the needed cockpit with all necessary information in only one screen!



A storage cockpit with BVQ helps to better answer the following questions

  • Are you getting the most out of your storage investment?
    • how is the capacity utilized?
    • what are the growth rates?
    • is storage performance as expected?

  • Do you have insight into the performance, utilization and health of your storage infrastructure?
    • is the storage tuned correctly?
    • is the storage infrastructure healthy ?
    • are there foreseeable problems which can impact services and revenue?
    • can you help yourself in case of storage problems
    • can your partner help in case of storage problems
    • when was the last regular health check - more than one month ago?

  • Have you balanced your capacity and performance requirements with the right storage tier?
    • is there a need or win to add new SSD or flash capacity?
    • how about adding some lower cost nearline storage?
    • does your easy tier perform well?
    • are there overload situations is a storage tier?
    • what is the future storage mix?

 

The CeBIT VersaStack / BVQ cockpit project

  • Target was to setup a complete monitoring system for the storage component of the IBM Cisco VersaStack solution which is a Storwize V7000 in this case.
  • For this prototype we used a small X220 laptop to setup the complete BVQ environment
  • This laptop was connected with ISCSI to the storage and used IOMeter to create some load on the storage. (this is the reason for the relatively high latencies in the picture)
  • Setup time was one hour 
  • Installation can now be improved by adding specific alerts or reporting with BVQ
 

Results

  • One screen dashboard
  • Enhanced with Storwize log information


    • Treemap / Treetable
      Monitor current load on the system - color codes will show storage overload situations
      Detail information about the performance of the single volumes using the Treetable.
      Enlarge and use the treemap for all kind of Analysis steps


    • CPU performance
      CPU load of all processors
      Max CPU core load to early detect overload in a CPU


    • Upper cache utilization
      Upper cache is the portal to the storage and every overload here leads to performance problems
      Enlarge and use this screen to find the volumes which cause the overload


    • Lower cache utilization
      Lower cache overflow is caused by backend storage problems and needs to be detected early
      Enlarge and use this screen to find out whether it is a backend or front end problem and dive deeper to find the root causes.


    • VDisk performance as overall performance from all VDisks
      Enlarge and use this screen as starting platform for every kind of volume related analysis
      You can easily add new indicators like transfer sizes here. you can drill down to the single disk or from the disks in any other layer like VDisk copy or MDisk , ports, supporting CPU and cores, ...


    • MDisk performance
      in this case we ar not only looking into a single MDisks we display all of them to see deviations in the performance of the single MDisks
      We could do the same for the drives but it makes sense to drill down into a drives from a conspicuous MDisk


    • Node port performance
      find overload or port problems like full buffers in this screen
      Resolve root causes of port problems by drilling down top the VDisks from a conspicuous port


    • Storwize Log information
      are visible in all single performance screens
      Whenever a change happens in the Storwize a new marker with some information will be placed into the screens


       

The CeBIT VersaStack / BVQ dashboard

 

Same picture with some comments