Skip to main content
NetApp Knowledge Base

How is read and write latency measured in Element Software ?

Views:
162
Visibility:
Public
Votes:
0
Category:
element-software
Specialty:
solidfire
Last Updated:

Applies to

  • Element Softweare
  • SolidFire ActiveIQ
  • writeLatencyUSecreadLatencyUSec parameters in Element APIs such as GetClusterStats, ListVolumeStats, GetVolumeStats

Answer

The timer of latency measured by Element Software start from receving SCSI Request. And the timer ends when the request completes. 

Here is more detail for each Read and Write I/O.

Read I/O

  1. Client sends a SCSI Read command
  2. Cluster receives the SCSI Read command
  3. Cluster injects QoS delay (if any)
  4. Cluster begins SCSI Read command processing (timer starts)
  5. Cluster reads data from the volume
  6. Cluster transfers read data to client
  7. If more data needs to be read, repeat step 5 and 6, otherwise the final transfer of read data includes SCSI status to complete the SCSI Read command (timer ends)
  8. Client receives the SCSI Status for the SCSI Read command

Write I/O

  1. Client sends a SCSI Write command with some/all write data
  2. Cluster receives the SCSI Write command
  3. Cluster injects QoS delay (if any)
  4. Cluster begins SCSI Write command processing (timer starts)
  5. If cluster needs more data to write, request data from the client
  6. Client sends additional write data to the cluster
  7. Cluster writes data to the volume
  8. If more data needs to be written, repeat step 5 through 7
  9. Cluster sends SCSI status to complete the SCSI Write command (timer ends)
  10. Client receives the SCSI status for the SCSI Write command

Note :

  • Latency includes the time for data transfer between client host and storage system. 
  • Exception is QoS delay such as reachinig MaxIOPS. The average latency observed by the client may differ from what is reported in the volume stats because of this.
  • These examples don't account for any transient I/O errors that would require a SCSI command to be retried by the client.

Additional Information

Active IQ collect stats every minute for last 0.5 sec of performance of data through management node.

 

NetApp provides no representations or warranties regarding the accuracy or reliability or serviceability of any information or recommendations provided in this publication or with respect to any results that may be obtained by the use of the information or observance of any recommendations provided herein. The information in this document is distributed AS IS and the use of this information or the implementation of any recommendations or techniques herein is a customer's responsibility and depends on the customer's ability to evaluate and integrate them into the customer's operational environment. This document and the information contained herein may be used solely in connection with the NetApp products discussed in this document.