ADSM-L

Re: Collocation Experience

1999-01-26 15:57:54
Subject: Re: Collocation Experience
From: Dwight Cook <decook AT AMOCO DOT COM>
Date: Tue, 26 Jan 1999 14:57:54 -0600
     V3 has multiple levels of collocation...
     no
     yes
     file
     Sounds like you might want collocation yes
     this would put individual clients on isolated tapes...
     tape selection for output (inbound backups/archives or migrations of
     diskpools or output during tape reclamation) would be
        1) a readwrite filling tape already containing data from that node
        2) a scratch tape
        3) any readwrite filling tape
     if you have collocation file the search is a little different in that
        1) blah containing data from client and from client's filespace
        2) scratch tape
        3) blah containing data from client
        4) any filling tape (any clients data)

     we have servers at all levels... no, yes, and file...  below is a list
     of our current environment's tape status... each line is a different
     adsm server (connected to a 3494 w/ 3590's) I removed the server names
     (security thing) but you can probably guess by looking at the numbers
     which servers are collocate no, yes, and file... there are 2 YES and 1
     FILESPACE, the rest are NO... OK I really should point out which is
     which but it is WAY TO EASY TO SPOT THEM... really.


Tue Jan 26 06:05:08 CST 1999 You have 46 SCRATCH and 314 PRIVATE volume s of
which 19 are filling, the rest are full.

Tue Jan 26 06:05:09 CST 1999 You have 26 SCRATCH and 610 PRIVATE volume s of
which 27 are filling, the rest are full.

Tue Jan 26 06:05:12 CST 1999 You have 80 SCRATCH and 708 PRIVATE volume s of
which 28 are filling, the rest are full.

Tue Jan 26 06:05:14 CST 1999 You have 36 SCRATCH and 307 PRIVATE volume s of
which 22 are filling, the rest are full.

Tue Jan 26 06:05:15 CST 1999 You have 21 SCRATCH and 403 PRIVATE volume s of
which 166 are filling, the rest are full.

Tue Jan 26 06:05:17 CST 1999 You have 136 SCRATCH and 687 PRIVATE volum es of
which 32 are filling, the rest are full.

Tue Jan 26 06:05:19 CST 1999 You have 270 SCRATCH and 179 PRIVATE volum es of
which 21 are filling, the rest are full.

Tue Jan 26 06:05:21 CST 1999 You have 185 SCRATCH and 73 PRIVATE volume s of
which 22 are filling, the rest are full.

Tue Jan 26 06:05:22 CST 1999 You have 91 SCRATCH and 9 PRIVATE volumes of which
14 are filling, the rest are full.

Tue Jan 26 06:05:23 CST 1999 You have 284 SCRATCH and 610 PRIVATE volum es of
which 22 are filling, the rest are full.

Tue Jan 26 06:05:27 CST 1999 You have 44 SCRATCH and 10 PRIVATE volumes
of which 18 are filling, the rest are full.

Tue Jan 26 06:05:28 CST 1999 You have 30 SCRATCH and 673 PRIVATE volume s of
which 522 are filling, the rest are full.

Tue Jan 26 06:05:31 CST 1999 You have 37 SCRATCH and 510 PRIVATE volume s of
which 247 are filling, the rest are full.




______________________________ Reply Separator _________________________________
Subject: Collocation Experience
Author:  shira (shira AT TLV.IEC.CO DOT IL) at unix,mime
Date:    1/26/99 7:39 AM


Hello all,

I need u to share your experience about working with collocation.

Lately I have ran some tested that showed that during restore, ADSM performs aro
und 14 mounts, which takes ~7 minutes each --> a fact that reflects the restore
time significantly.

Now we consider working with collocation. My site contains 150 clients, most of
the clients r relatively small (q auditocc --> ~10 Gb each), and few clients (~2
0) occupies large volume of storage (q auditocc --> ~120 Gb each)

I read the archive list, and found the discussion about the tape wear while work
ing with collocation. The data I wish to know is:

 * During time, does ADSM keeps the data of a client on minimum number of tapes,
 or does the efficiency of the collocation procedure drops down?
 * Does the migration process finishes on time (before new round of backups) whi
le collocation is set on.
 * What is the average amount of tapes that contains one client data.



Any input will be welcomed.



Best Regards,


Shira.
<Prev in Thread] Current Thread [Next in Thread>