How to collect WAFLTOP output from CLI

Last updated
Save as PDF
Share
1. Share
2. Tweet
3. Share

Views:: 3,172

Visibility:: Public

Votes:: 4

Category:: ontap-9

Specialty:: perf

Last Updated:

Applies to

Data ONTAP 8 7-Mode
ONTAP 9

Description

wafltop is used to collect low-level WAFL performance data on a per-volume basis.
It may be used if using 7-mode (in place of qos statistics) or if qos statistics cannot help identify CPU or disk utilization latency sources.
It tracks utilization of the following resources:
- CPU
- Disk I/O
- Memory
- NVLOG utilization
The format of the volume name will be: aggregate:volume:protocol:message in the out:
- aggregate is specifically the name of the Aggregate the Volume resides in.
- volume is the name of the FlexVol.
- protocol refers to frontend CIFS, FCP, iSCSI, or NFS, or backend protocols like spinvfs, dense, walloc, zombie, or other.
  - walloc is write allocator
  - dense refers to deduplication/storage efficiency
  - zombie refers to deletions
  - other is a catchall for any other internal operations, such as WAFL scanners
  - spinnp refers to Clustered/ONTAP 9 protocol workloads
  - There may be others but these will cover any situation that is likely to be encountered.
- message is an internal WAFL message, but is only meant to be used by NetApp Employees only
  - There may be a few things that can be seen such as:
    - backdoor refers to some internal WAFL operations that may not be classified in other operation types, but are background in nature
      - Example: A FlexClone is being split
    - file_io refers to anything passing to/from disk
    - SHARING* refers to deduplication (scheduled/post-process or inline)
- If a volume has a frontend protocol infront of it, then user workload is referred to.

Terminology:

CPU Utilization - Collected for kahuna, wafl_exempt, xcleaner domains. CPU utilization refers to the percentage of CPU cycles used.
Latency histograms - System (internal) and client latency.
Disk I/O operations - Reads and writes from disk in MB and in terms of RAID messages.
NVLog utilization - NVLog utilization for normal Consistency Points (CPs) and back-to-back CPs (b2b CPs). NVlog utilization is shown in terms of average percentage of the log used per CP.
Memory utilization(Debug only) - shows buffer cache and inode cache usage in terms of MB used.
- Note: Not meant to be used for performance troubleshooting, but only by NetApp Engineering.
I/O Utilization- I/O utilization refers to the amount of data read or written in MB/s.