Hi Stephen. We have ~450 tb of isilon, and we bought a solution from
Fuji called dternity to help with archiving.
http://www.dternity.net/
It's basically a NAS with a disk buffer + tape drives + tape library
that send the data to LTFS formatted tapes.
It has some terabytes of disk storage built in which it splits into read
and write cache, (which is expandable via FibreChannel disk like a
nexsan array).
The write cache eases ingest, so you are not writing directly to tape,
you can nfs mount it and just copy or rsync the data on to it.
You can set policies on how long the data has to be in the cache before
it writed it to tape, and how long to retain the disk copy before it is
flushed(you can write to tape after 1hr in the cache, but you could keep
it in the cache for a day if you wanted to). The metadata I believe is
retained on disk(so ls's against it don't incur tape reads), and there
is a setting for leaving a stub file of a preset size(globally, not per
share), so if a client requests a file, while a tape is being loaded,
the client can be fed some of the stub data from disk. I think their
sweet spot in the market is large media files, like 4k video, but we
have found it works ok for some of datasets that our users have
generated and want to keep but don't need to be wasting the expensive
space on the isilon to keep. We allow users to retrieve the files
themselves(but we migrate the data into the archive for them), and let
them know that if it's on the archive nfs mount that it will not have
isilon level performance, and if they need that they need to copy it
back onto the isilon.
If you're interested in hearing more about it, let me know, and I'll see
if I can get you and one of our storage guys together to talk about it.
rgt
On 05/04/2015 12:18 PM, Stephen B Goldman wrote:
Hello list members,
We currently have Isilon NAS with 600 TB of data. We
find we have 20 to 40 Tb of lab data we would like to offload spinning
disk. We have TSM available from the campus IT group. They offer an
archiving service.
We would like to hear from others organizations that
have the same need and listen to how they are handling this issue.
Thanks in advance,
Stephen
Stephen Goldman
Systems Administrator
Massachusetts Institute of Technology
Department of Biology
31 Ames Street 68-211
Cambridge, Ma 02139
[email protected] -617-452-2595
_______________________________________________
bblisa mailing list
[email protected]
http://www.bblisa.org/mailman/listinfo/bblisa
_______________________________________________
bblisa mailing list
[email protected]
http://www.bblisa.org/mailman/listinfo/bblisa
