Analysis of performance concerns and fine-tuning

Information on addressing performance problems due to the high number of active concurrent SMB connections on each protocol node of IBM Storage Scale.

If IBM Storage Scale experiences performance problems, the following actions can be taken:

• You can use the IBM Storage Scale center GUI or CLI to understand which physical resource such as CPU, memory, networking, disk in the system are highly used resources. This information helps to gain an insight on the physical system resource that is possibly inhibiting or limiting performance. For more information, see mmperfmon command.
• Ensure that the IBM Storage Scale protocol nodes are configured with the maximum number of processors, memory, and networking adapters.
• Add more IBM Storage Scale protocol nodes to your system.
• Move certain advanced functions, such as TSM and AFM, to periods of time when SMB file activities that are related to the server is lower.
• Reduce the frequency at which snapshots are being created and deleted either at the file system or the file set level or both, especially during the periods of highest SMB user activity.
• Check that the file system cluster has appropriate values for pagepool, number of worker threads, and number of receiver threads. For more information, see Configuring and tuning your system for GPFS.
• Investigate and fine-tune the performance of the underlying disk storage systems that contain the file systems on which the SMB export resides. The checklist items are described in the following list:
  • Ensure that the file system disks of a storage system are distributed between the pair of IBM Storage Scale storage nodes to which that storage system is attached. One half of the file system disks in a storage system should have one of the storage nodes that are identified as the primary NSD server. The other half of the file system disks in the storage system should have the other IBM Storage Scale storage node in the pair that is assigned as the primary NSD server. The mmlsdisk CLI command option shows the IBM Storage Scale storage nodes that are the primary and secondary NSD server for each file system disk.
  • If an IBM Storage Scale metadata replication or data replication is being used, ensure that you assign file system disks to failure groups that balance the I/O and data across a set of file system disks, RAID arrays, and disk storage systems. The mmlsdisk CLI command shows the failure group to which each file system disk is assigned. The mmchdisk CLI command can be used to change the failure group to which each file system disk is assigned.
  • If the underlying disk storage systems on which the file system reside is becoming a performance bottleneck, consider adding more physical resources to those disk storage systems. More resources include more cache memory, more disk drives, more RAID arrays and more file system disks for the file systems that contain the SMB exports.
  • If the existing disk storage systems on which the file system resides reach their limit in terms of either capacity or performance or both, then consider adding more disk storage systems and extending the file system on which the SMB exports resides. Capacity and performance improvement can be done by adding new file system disks that are residing on the new disk storage systems to the existing disc storage system.