Questions on recent Volume Performance Metrics change #4280
-
|
Hello, I have couple of questions as I'm trying to better understand the volume performance metrics, including the changes recently introduced at #3900 Here we upgraded Harvest from 25.08 to 26.02:
As per above screenshot, READ and WRITE seem to have gotten a rise, more significant for write latency, while OTHER latency seem to report less. 
According to the 25.08 docs, the source was
and afterwards just:
So shall I just assume that unlike before, now WRITE latency also includes request processing and network communication time? And it's exactly the same for READ.
On #3900 it is mentioned:
So, I would expect that they are consumed from the same (QoS) endpoint, however it is not the case.
Theoretically, what is then the difference in matter of actual values they display between volume metrics (volume_read_ops, volume_write_ops, etc.) and qos volume metrics (qos_read_ops, qos_write_ops, etc.). 
Many thanks in advance for your input!! |
Beta Was this translation helpful? Give feedback.
Replies: 1 comment 4 replies
-
|
@spapadop You are correct. changes in values are expected with this update, and the reasons are explained in the KB article at https://kb.netapp.com/on-prem/ontap/Perf/Perf-KBs/What_is_the_difference_in_latency_of_the_same_vol…. The |
Beta Was this translation helpful? Give feedback.
-
|
Many thanks! That makes my second question clear. Regarding the first one, I still don't get what's the actual difference, i.e., what is currently measured as part of READ and WRITE latency. Is it whatever it was measured before + request processing time + network communication time ? And those two were taken away from OTHER and moved into READ + WRITE? |
Beta Was this translation helpful? Give feedback.
-
|
The differences from ONTAP are documented here: This explains that, for NAS, the total operations do not equal the sum of read, write, and other operations when using Raising an ONTAP ticket may help provide more detailed information about these APIs. |
Beta Was this translation helpful? Give feedback.
-
|
It may be that I don't get the KB, but I'm not asking about "Total Ops" and if it matches the sum of READ, WRITE, OTHER or not. Let's target my question specifically on 
Theoretically we moved from One plausible answer would be that unlike before harvest upgrade, write latency now also includes request processing and network communication time. Unfortunately I don't have internal ONTAP knowledge or exposure to dig out the truth. If that question is better suited for the netapp support, let me know and I'll open a case accordingly, thanks! |
Beta Was this translation helpful? Give feedback.
-
|
@spapadop Yes, this question is best addressed by the ONTAP team. Please submit a NetApp support ticket for further assistance. Thanks. |
Beta Was this translation helpful? Give feedback.
@spapadop You are correct. changes in values are expected with this update, and the reasons are explained in the KB article at https://kb.netapp.com/on-prem/ontap/Perf/Perf-KBs/What_is_the_difference_in_latency_of_the_same_vol….
The
api/storage/volumes?fields=statisticsendpoint is consistent withapi/cluster/counter/tables/qos, and both provide the same statistics. Therefore, the performance metrics labeled as qos_ and volume_ will have similar values after this change. You might see slight differences because Harvest polls qos metrics every 3 minutes and volume performance metrics every 1 minute.