Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!ucbvax!ICAEN.UIOWA.EDU!dbfunk
From: dbfunk@ICAEN.UIOWA.EDU (David B Funk)
Newsgroups: comp.sys.apollo
Subject: Re: omniback
Message-ID: <8909081030.AA00444@icaen.uiowa.edu>
Date: 8 Sep 89 09:58:18 GMT
Sender: daemon@ucbvax.BERKELEY.EDU
Organization: Iowa Computer Aided Engineering Network, University of Iowa
Lines: 41

WRT posting from achille%cernvax%mcvax.uucp@uunet.uu.net  (achille petrilli)

> On a, more or less, related argument, I'd like to know if somebody has tested/used
> the new backup facility from Apollo, NBS Omniback. I'm interested in hearing any
> quirks/problems you had and about its performance.

Here's my first pass opinion of Omniback:
    Its fast, its neat, but its a little rough around the edges.

Specifics:
    OS: sr10.1, Omniback v1.0, Exabyte attached to DN3500 SCSI port.

Up side:
    The concept seems sound. The "work list" and scheduler are usefull
    powerfull tools. The dialog interface provides usefull progress
    information. It is fast, doing a "-conc 3" from a bunch of DN3500s
    with 348 Meg disks, I saw about a 100Kbyte/sec backup rate.

Down side:
    It doesn't have some of the nice features that Workstation Solutions
    "tbs" product has, such as the index listing depth control.
    It seems rather fragile. Due to simple problems, that wouldn't have
    caused wbak to even pause, the Omnibackup of a complete disk crashed.
    IE the data reader process on the node died so that the backup of that
    disk was aborted. The backup of the other disks in progress wasn't
    affected, but the offending disk wasn't backed up. But, out of 8 disks
    in my backup test, the backup failed on 3 of them. Simple things like
    an abandoned floppy disk mount point caused the failures. These were
    things that would cause wbak to report an "object not found" error,
    but it would continue to back up the rest of the disk. When a backup
    on a disk failed, it would leave behind a "data reader" process that would
    then go into an infinite loop and eat up all spare CPU cycles until
    killed by hand.

This is a brand new product and the first version, so a few bugs are to
be expected. I've only given it a few tests and havn't tried all modes
of operation. In non-concurrent mode it may work better.
Once enough bugs are worked out so that it's more reliable,
it'll be a good tool. For now though, we are still using wbak.

Dave Funk