Unsolved
This post is more than 5 years old
14 Posts
0
1106
Is there a way to get the PRECISE "Used Object Count" ?
We would like to compare a "embedded" blob archiving process
to an "linked" blob archiving process
due to the used Object Counts after e.g. archiving 1 million
of CClips.
But displaying the used Object Count via
"show capacity total"
just shows
Used Object Count: 36 M
"show capacity availability"
would just show
Used Object Count: 36 M (18%)
Now 36M is quite unprecise to make a decision between embedding vs. linking.
Our question is:
Is there a way to get the PRECISE "Used Object Count" ?
cincystorage
2 Intern
2 Intern
•
467 Posts
0
July 5th, 2013 05:00
"show pool capacity" should give you the level of detail you're looking for..
mckeown_paul
409 Posts
0
July 5th, 2013 05:00
I had a quick look at the documentation and I couldnt see anything that would give you the precise object count, although the centera health report may give you that (I couldnt find an example to check).
However if you know what your object sizes are before you ingest them and you set the embedded threshold to 100KB (the max) then you can figure out the numbers that were embedded or not.
wmluecke
14 Posts
0
July 5th, 2013 05:00
Thanks.
Yes we see that embedding is working.
The opened cclips in cclip browser show a base64 embedded data and the graphical view would show that too.
But we do use different size of blobs which can sometimes not be embedded.
Thus we would like to compare the used object count of different storage strategies
when a specific representative amount of data is stored.
Therefore the precise used object count number would be extremely helpful.
wmluecke
14 Posts
0
July 5th, 2013 05:00
show pool capacity shows e.g.
Capacity / Pool Quota Used Free C-Clips Files
------------------------------------------------------------------------------
windream 98 GB 20 GB 77 GB 147870 295973
where Files is NOT the "used object count".
mckeown_paul
409 Posts
0
July 5th, 2013 05:00
Note that if you embed you will use only 2 centera object instead of 4.
You can confirm this if you suspect it doesnt work as documented by using jcasscript to open a clip and use the cliprawview to look at the format of the CDF. If the embedded blob functionality is not working you'll see only about a screens worth of text. If it does work as documented you see screens worth of text as the blob will be shown as a tag attribute that has been base64 encoded.
wmluecke
14 Posts
0
July 5th, 2013 06:00
thanks for your effort.
now to avoid some answers to our customers like "you may probably avoid about some million object counts
when using the embedded strategy but we do not know exactly"
could you answer following questions:
a) how many object counts are used for
1 clip with 3 files embedded
1 clip with 2 files embedded + 1 file linked
1 clip with 2 files embedded + 0 file linked
1 clip with 1 file embedded and 2 files linked
1 clip with 1 file embedded and 1 file linked
1 clip with 1 file embedded and 0 files linked
1 clip with 3 files linked
1 clip with 2 files linked
1 clip with 1 file linked
b) then additionally the same questions with the option that 1 linked file is identical for 1 or more other clips
so that single instance storage is used.
c)
all questions above when all clips are replicated .
d) and one precise question:
how many object counts are used for following situation (note that there are more than this one pool configured):
pool C-Clips Files
windream 155394 310788
when all cClips are saved with 2 embedded files.
where show capacity availabilityshows:
Tottal Object Count: 200 M (100%)
Used Object Count: 36 M (18%)
Free Object Count: 164 M (82%)
mckeown_paul
409 Posts
1
July 5th, 2013 07:00
I'll not answer every individual question but hopefully give you the information so you can answer them yourself
The CDF will use up 2 centera objects regardless of how many blobs/files are embedded.
Each linked blob with use up 2 centera objects if CPM is used or 7 if CPP.
If there are linked files that are identical and you are using storage policy capacity then only one copy of the linked blob is stored and the multiple CDF's reference it. When you start single instancing like this it's impossible to extrapolate the object count used from the number of files ingested. We have no tools to show you the number of objects that are being referenced by multiple CDF's.
When clips are replicated they will consume the same number of object on the replica target assume both use the same protection mechanism (CPM or CPP) and both have the same storage policy set (i.e. SIS is enabled on both).
wmluecke
14 Posts
0
July 5th, 2013 07:00
Calculating the Used Object Count by the above method seems to be the only way to get a DELTA between
different archive strategies.
Also the health report would not really show the precise "Used Objects" counter.
1.
show report health c:\centera_health.xml would get the xml data section:´
usedObjectCount="36449530"
totalObjectCount="200000000"
freeObjectCount="163550470"
changes after archiving 2 C-clips, one of them with 1 and the other with 2 linked (NON-EMBEDDED) files:
usedObjectCount="36450089"
totalObjectCount="200000000"
freeObjectCount="163549911"
Note that the difference of 559 has nothing to to with the real archived clips and files.
2.
Also the pool capacity object count would just show exact these file-counters which can NOT be the object Counts:
usedCapacity="22216708400"
usedObjectCount="391083"
usedClipCount="195206"
changes after archiving 2 C-clips, one of them with 1 and the other with 2 linked (NON-EMBEDDED) files:
usedCapacity="22216720904"
usedObjectCount="391086"
usedClipCount="195208"
Conclusion:
the "real" precise Used Object count can NOT be retrieved by health reports.