When troubleshooting performance issues, the very first point is to frame the problem. This will save you a lot of time. Here are the good questions to raise.

  • WHO
    • Client access issue ?
    • ONE client or ALL clients ?
    • ONE data pool or all data pools
    • ONE share or all shares
    • ONE FC/iSCSI volume or all volumes
  •  WHAT
    • What performance issue are you seeing ?
    • What performance are you expecting to see ?
    • How does the performance issue manifest itself ?
      • low network throughput/high network latency
      • low disk activity
      • high disk latency
      • high CPU load
      • high iSCSI|NFS|CIFS|FC|ftp Latency
      • etc …
    • Provide details of any appliance functionality in use
      • replication ?
      • snapshot schedule ?
      • dedup – note, since 8.7.0 dedup V2 has been introduced and improves performance a lot
      • compression ?
  • WHEN
    • Certain times of the day ?
    • Continuous or periodic/intermittent, for how long ?
    • After changes to hardware/configuration
    • After firmware upgraded ?
    • After functionality enabled (snapshots, dedup etc.) ?
  • WORKLOAD DETAILS
    • What protocols are being used (NFS/CIFS/iSCSI/FC/Infiniband) ?
    • Is it a predominantly READ workload ?
    • Is it a predominantly WRITE workload ?
    • Is it a mixed READ/WRITE workload ? (approximate mix ?)
    • Is it a RANDOM access workload ?
    • Is it a SEQUENTIAL access workload (backups, RMAN)
    • number of clients
    • client configuration (including operating system version)
    • Virtual machines : Vmware, VDI, …
    • If Virtual machines, which OS : Windows, Linux, Solaris …
    • ORACLE Database ? Other Database ?