Automatic Storage Management (ASM) metrics are a goldmine: Let's use them!

Automatic Storage Management
(ASM) Metrics are a Goldmine:
Let’s Use Them!
Bertrand Drouvot

About Me
 Oracle DBA since 1999
 OCP 9i,10g,11g
 Rac certified Expert
 Exadata certified implementation specialist
 Blogger since 2012
 @bertranddrouvot
 BasketBall fan

Are you happy with?
 asmcmd iostat?
 asmiostat.sh from MOS [ID 437996.1])?
 Me, I am not: The metrics provided are not
enough, the way we can extract and display
them is not customizable enough, and we
don’t see the I/O repartitions within all the
ASM or database instances into a RAC
environment.

Welcome to asm_metrics
1. It provides useful real-time metrics:
 Reads/s: Number of read per second.
 KbyRead/s: Kbytes read per second.
 Avg ms/Read: ms per read in average.
 AvgBy/Read: Average Bytes per read.
 Writes/s: Number of write per second.
 KbyWrite/s: Kbytes write per second.
 Avg ms/Write: ms per write in average.
 AvgBy/Write: Average Bytes per write.
2. It is RAC aware: You can display the metrics for all the ASM and (or)
database instances or just a subset.
3. You can aggregate the results following your needs in a customizable way:
Aggregate per ASM Instances, database instances, Diskgroup, Failgroup or
a combination of all of them.
4. It does not need any change to the source: Simply download it and use it.

How does it work?
 The script takes a snapshot each second (default interval)
from the gv$asm_disk_iostat cumulative view (or
gv$asm_disk_stat) and computes the delta with the
previous snapshot.
 The only difference with gv$asm_disk_stat is the
information available in memory while v$asm_disk access
the disks to re-collect some information.
 Since the information required doesn’t require to “re-collect”
it from the disks (as a discovery of new disks is
not needed), gv$asm_disk_stat is more appropriated here.

Let’s use it
Important remark:
The blank value for one of those fields (INST, DBINST, DG, FG, DSK) means that the values have
been aggregated for this particular field.

How are the metrics computed?
 The metrics are computed this way:
 Reads/s comes from the delta computation of the READS column divided by the
snapshot wait interval.
 KbyRead/s comes from the delta computation of the BYTES_READ column divided by
the snapshot wait interval.
 Avg ms/Read comes from the delta computation of the READ_TIME / READS columns.
 AvgBy/Read comes from the delta computation of the BYTES_READ / READS columns.
 Writes/s comes from the delta computation of the WRITES column divided by the
snapshot wait interval.
 KbyWrite/s comes from the delta computation of the BYTES_WRITTEN column divided
by the snapshot wait interval.
 Avg ms/Write comes from the delta computation of the WRITE_TIME / WRITES
columns.
 AvgBy/Write comes from the delta computation of the BYTES_WRITTEN / WRITES
columns.

What are the features? (1/3)
 To explain the features, let’s have a look to
the help

1. You can choose the number of snapshots to display and the time to wait between the
snapshots. The purpose is to see a limited number of snapshots of a specified amount of
wait time between snapshots.
2. You can choose on which ASM instance to collect the metrics thanks to the -
INST= parameter. Useful in RAC configuration to see the repartition of the ASM metrics
per ASM instances.
3. You can choose for which DB instance to collect the metrics thanks to the -
DBINST= parameter (wildcard allowed). In case you need to focus on a particular
database or a subset of them.
4. You can choose on which Diskgroup to collect the metrics thanks to the -DG= parameter
(wildcard allowed). In case you need to focus on a particular diskgroup or a subset of
them.
5. You can choose on which Failgroup to collect the metrics thanks to the -FG= parameter
(wildcard allowed). In case you need to focus on a particular failgroup or a subset of
them.

6. You can choose on which Exadata Cells to collect the metrics thanks to the -
IP= parameter (wildcard allowed). In case you need to focus on a particular cell or a
subset of them.
7. You can aggregate the results on the ASM instances, DB instances, Diskgroup, Failgroup
(or Exadata cells IP) level thanks to the -SHOW= parameter. Useful to get an overview of
what is going on per ASM Instances, per diskgroup or whatever you want, as this is fully
customizable.
8. You can display the metrics per snapshot, the average metrics value since the collection
began (that is to say since the script has been launched) or both thanks to the -
DISPLAY= parameter. So that you can get the metrics per snapshots, since the script has
been launched or both.
9. You can sort based on the number of reads, number of writes or number of IOPS
(reads+writes) thanks to the -SORT_FIELD= parameter (so that you could for example
find out which database is the top responsible for the I/O). So that you can find the ASM
instances, the database Instances, or the diskgroup, or the failgroup or whatever you
want that is generating most of the I/O reads, most of the I/O writes or most of the IOPS
(reads+writes).

Find out the most physical IO consumers through
ASM in real time. This is useful as you don’t need to
connect to any database instance to get this info as
this is "centralized" into the ASM instances.
Let's sort first based on the number of reads per
second that way:
./asm_metrics.pl -show=dbinst -sort_field=reads
11
Use case 1

I want to see the ASM preferred read in action for a particular diskgroup
(BDT_PREF for example) and see the IO metrics for the associated
failgroups. I want to see that no reads are done "outside" the preferred
failgroup.
Let’s configure the ASM preferred read parameters:
SQL> alter system set asm_preferred_read_failure_groups='BDT_PREF.WIN' sid='+ASM1';
System altered.
SQL> alter system set asm_preferred_read_failure_groups='BDT_PREF.JMO' sid='+ASM2';
System altered.
And check its behaviour thanks to the utility:
./asm_metrics.pl -show=dg,inst,fg -dg=BDT_PREF
12
Use case 2

I want to see the IO distribution on Exadata across the Cells
(storage nodes). For example I want to check that the IO
load is well balanced across all the cells. This is feasible
thanks to the show=ip option:
./asm_metrics.pl -show=dbinst,dg,ip -dg=BDT
13
Use case 3

I want to see the IO distribution recorded into the ASM instances:
./asm_metrics.pl -show=inst
I want to see the IO distribution recorded into the ASM instances for each
database instance:
./asm_metrics.pl -show=inst,dbinst
I want to see the IO distribution recorded into the ASM instances for the
database instances linked to the BDT database:
./asm_metrics.pl -show=inst,dbinst -dbinst=%BDT%
14
Use case 4, 5 & 6

I want to see the IO distribution over the FAILGROUPS:
./asm_metrics.pl -show=fg
I want to see the IO distribution and their associated metrics across the
ASM instances and the failgroups:
./asm_metrics.pl -show=fg,inst
I want to see the IO distribution across the ASM instances, diskgroups and
failgroups:
./asm_metrics.pl -show=fg,inst,dg
15
Use case 7,8 & 9

The use cases focused only on snapshots taken during the last second
but you could also:
Takes snapshots of longer period of time thanks to the
interval parameter:
./asm_metrics.pl -interval=10 (for snaps of 10 seconds)
View the average since the collection began (not only the
snaps delta) thanks to the display parameter that way:
./asm_metrics.pl -show=dbinst -sort_field=iops -display=avg
16
Remark

For this I created the csv_asm_metrics utility to produce a csv file from the output
of the asm_metrics utility.
Once you get the csv file you can graph the metrics with your favourite
visualization tool (I’ll use Tableau as an example).
First you have to launch the asm_metrics utility that way (To ensure that all the
fields are displayed):
-show=inst,dbinst,fg,dg,dsk for ASM >= 11g
-show=inst,fg,dg,dsk for ASM < 11g
and redirect the output to a text file:
./asm_metrics.pl -show=inst,dbinst,fg,dg,dsk > asm_metrics.txt
17
Graphing ASM metrics

Produce the csv file
./csv_asm_metrics.pl -if=asm_metrics.txt -of=asm_metrics.csv -
d='2014/07/04'
The csv file looks like:
Snap Time,INST,DBINST,DG,FG,DSK,Reads/s,Kby Read/s,ms/Read,By/Read,Writes/s,Kby Write/s,ms/Write,By/Write
2014/07/04 13:48:54,+ASM1,BDT10_1,DATA,HOST31,HOST31CA0D1C,0,0,0.0,0,0,0,0.0,0
2014/07/04 13:48:54,+ASM1,BDT10_1,DATA,HOST31,HOST31CA0D1D,0,0,0.0,0,0,0,0.0,0
2014/07/04 13:48:54,+ASM1,BDT10_1,DATA,HOST32,HOST32CA0D1C,0,0,0.0,0,0,0,0.0,0
2014/07/04 13:48:54,+ASM1,BDT10_1,DATA,HOST32,HOST32CA0D1D,2,32,0.2,16384,0,0,0.0,0
2014/07/04 13:48:54,+ASM1,BDT10_1,FRA,HOST31,HOST31CC8D0F,0,0,0.0,0,0,0,0.0,0
2014/07/04 13:48:54,+ASM1,BDT10_1,FRA,HOST32,HOST32CC8D0F,0,0,0.0,0,0,0,0.0,0
2014/07/04 13:48:54,+ASM1,BDT10_1,REDO1,HOST31,HOST31CC0D13,0,0,0.0,0,0,0,0.0,0
As you can see:
1.The day has been added (to create a date) and next ones will be calculated (should the snaps 18

19
Visualize top IO consumers

21
Read IO distribution by Failgroup

22
Should I use the ASM preferred read?

Without (Read IOPS and Throughput)
23
Simulate the ASM preferred read (1/3)

Create calculated field
24

With simulated FG
25

 Thanks to these use cases, I hope you can see how
customizable the utility is and how you could take
benefit of it in a day-to-day work with ASM.
 The main entry for the tool is located to this blog
page:
http://bdrouvot.wordpress.com/asm_metrics_scri
pt/ from which you’ll be able to download the
script or copy the source code.
 Feel free to download it and to provide any
feedback.
26
Conclusion

Automatic Storage Management (ASM) metrics are a goldmine: Let's use them!

More Related Content

What's hot (17)

Similar to Automatic Storage Management (ASM) metrics are a goldmine: Let's use them! (20)

Recently uploaded (20)

Automatic Storage Management (ASM) metrics are a goldmine: Let's use them!