home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!ogicse!uwm.edu!fps.mcw.edu!post.its.mcw.edu!not-for-mail
- From: kcb@post.its.mcw.edu (Kent C. Brodie)
- Newsgroups: comp.os.vms
- Subject: Re: Bound volume sets: are they a bad idea?
- Message-ID: <1egbrqINNiqd@post.its.mcw.edu>
- Date: 19 Nov 92 03:28:58 GMT
- Article-I.D.: post.1egbrqINNiqd
- References: <1992Nov18.122133.1@rhodes.aero.org>
- Organization: Medical College of Wisconsin (Milwaukee, WI)
- Lines: 74
- NNTP-Posting-Host: post.its.mcw.edu
- X-Newsreader: TIN [version 1.1 PL6]
-
- singer@rhodes.aero.org wrote:
-
- : >>As far as I can reason it out, the pro is some degree of load balancing
- : >>between the disks, and no management hassles sorting out what to put where.
- : >>The con is a halved MTBF on the bound set. Backup shouldn't be a problem,
- : >>both together will fit on one DAT.
-
- Wrong. Volume binding does NOT do much load balancing at all. At the
- very least, you have absolutely no control over it. My best example
- is our own shop.... we have humungous MUMPS.DAT files (on the
- order of several million blocks large), and we used to have bound
- sets. Most of the time, the first (n-1) disks were totally idle,
- while the last disk in the set got absolutely pounded. Your
- mileage will vary, of course.
-
-
- : >>Would anyone care to tell me what I've missed, or of any experiences (bad
- : >>or long-term OK) with a volume set like the one I'm considering?
-
- long term? NOT a real good idea, unless you can live with the
- failure of one of the members, which effectively trashes the entire
- set. It all depends on your data. We *had* to go to bound
- sets because of the size of our files. (HUGE). However,
- the "importance" of our data, and our system performance needs
- both rose enough, where we now use disk STRIPING (works damned well,
- in my opinion, and well worththe money), and we also use volume
- shadowing. ($$$$$$$, but again, worth it for the peach of mind.
- And we *have* had enough failures to warrant it.)
-
- In our volume-binding days, we lost 3 disks in five years, but
- each of those losses were "fatal" and unrecoverable, except for
- using the previous night's backup.
-
- Could you live with that? If so, then binding is fine...
-
- : Cluster factor DOES not increase! The cluster factor is whatever must be at
- : initialize time ($INIT/CLUST=x) which is not a function of the number of
- : devices, but rather the size of the device. Once you bound the volumes,
- : you can verify this by doing $ SHOW DEVICE/FULL and also a $DIR/SIZE=ALL
- : You will see that they are both the same.
-
- This is correct. When adding VOLUMES, the cluster size does not
- increase. However, a STRIPE SET is different. It's not reeally
- volumes, but it *is* considered one extremely large SINGLE VOLUME.
- That's why the cluster size is larger, because of the SINGLE VOLUME
- size. This is also why stripe sets can only be 8.5 GB large.
- That's the largest addressable space in VMS for a SINGLE disk volume.
-
- With binding, however, you can get really really huge. (if you want..)
-
-
- : The simplicity of growing the set by simply adding more devices, and the
- : backing up of a single logical device overrides the other considerations.
-
- This is ONE area where binding had striping licked... in a bound set,
- it's a trivial operation to add a disk to the set. In striping,
- however, you have to back up your data, dissolve the stripe set,
- and re-build it from scratch (adding the new member, of course).
- If you have volume shadowing (as we do), this isn't a real big deal,
- but it is certainly time-consuming...
-
- I've done both volume binding and disk striping and volume shadowing
- (phase I and phase II) for years, so feel free to ask me if you
- have any more questions.
-
- --------------
- Kent C. Brodie - Sr. Systems/Network Manager brodie@post.its.mcw.edu
- Information Technology Systems
- Medical College of Wisconsin (414) 257-8769
- --
- --------------
- Kent C. Brodie - Sr. Systems/Network Manager brodie@post.its.mcw.edu
- Information Technology Systems
- Medical College of Wisconsin (414) 257-8769
-