MSA 1500 show perf command

[Max Sorokin]

Hi All,

Can somebody describe me the output values of the show perf command. The values are: LWrites LReads AvQDep CHits CMiss RACount LSectRd LSectWr


In attachment you can see the output of this command, and what i need to know i mark with a red color.



Thanks in Advance,
BR,
Max

[IBaltay · Accepted solution]

Hi,
the counters explanation is the following:

LWRITES
Number of logical write requests submitted by the host (hexadecimal)

LREADS
Number of logical read requests submitted by the host (hexadecimal)

AVGQdep
Average number of logical read and write requests queued. (decimal)

CHITS
Number of cache hits (host-generated logical read requests which
could be serviced entirely by the cache without any disk access)(hexadecimal)

CMISS
Number of cache misses (hexadecimal)

RACOUNT
Number of sectors pre-fetched (hexadecimal)

LSECTRD
Number of logical blocks read from this volume by the host.
Does not include any internally generated read requests such
as read-ahead, capacity expansion, etc. (hexadecimal)

LSECTWR
Number of logical blocks written to this volume by the host.
Does not include any internally generated write requests such
as cache flush, capacity expansion, etc.
(hexadecimal)

[Max Sorokin]

Thanks a lot, just another question:

What does it mean this terms:

Sample interval:

Command count:

Avg Command Latency:

% Time Active Reqs:

Max Outst Cmd Lists:

Logical Req Count:

Avg Locked Stripes:

Avg Stripes Waiting:

Avg Completions/Sec:

Avg Log Req Comp/Sec:


BR,
Max

[IBaltay]

Hi,

Sample interval:
Time since perf stats were last cleared

Command count:
Number of command lists submitted by the host

Avg Command Latency:
The average amount of time from when the command arrives
at the controller to its completion.

% Time Active Reqs:
% of time at least one logical request is outstanding, including event notification commands

Max Outst Cmd Lists:
Maximum number of outstanding commands lists

Logical Req Count:
Number of logical requests submitted by the host

Avg Locked Stripes:
The average number of stripes that are locked. (a stripe
is a group of physical sectors across a RAID group that are busy performing a write operation (other operations to the same sectors are locked out during this process).


Avg Stripes Waiting:
The average number of requests that are stalled waiting to access stripes that are locked.
A large value indicates random writes over a small area of a RAID 5 array.

Avg Completions/Sec:
The average I/O completed per second (IOPS

Avg Log Req Comp/Sec:
The average logical requests completed per second

[Max Sorokin]

Thanks a lot :)