Demetrius,
IMHO this is very cheap and too much shifted report with nothing analytical
in it. Set of coloured bars with different tags on them. Not good even for
advertisement. Of course you have the right to like it. But it even does
not answer to the initial thread question.
This might be called "analysis" somehow but is it "nice" ???
1. Is there an administrator achieving only 0.6 MB/s on a DLT7000 drive
being not sure this system badly needs tuning. Backup to disk is more than
two times slower than backup to tape (10.5 MB/s vs 4.7 MB/s)??? And who
performed the tuning "using IBM/Tivoli redbooks and product manuals" ??
This same Veritas Support Engineer?
2. "NetBackup v3.2 and Storage Manager v3.7.2 are enterprise class storage
applications" !!! However the tests performed are tasks for standalone
backup tools like integrated Windows NT Backup for example and there is
*NOTHING* enterprise class. And if using NT Backup on those test I achieve
higher throughput does this automatically mean I can backup very quickly
600 GB Oracle database residing on AIX, HP-UX or Solaris to this same NT
box ?!?
- not tested backup to more than one drive in parallel!
- not tested backup over LAN !!!
- not tested backup of many boxes to one server simultaneously !
- not tested repeated backups with many or only few files changed
(remember incremental forever) !
- not tested platform interoperability - backup of NT box to HP-UX
server or AIX database to NT server, etc. !
- not tested or even mentioned backup of applications - RDBMS (DB2,
MS SQL, Oracle, Sybase), Mail (Domino, Exchange), etc. !
- not tested online backups - neither open files or running
applications !
- not tested point-in-time or any other multi-version restore
(there was thread on this list where last night backup was virus infected)
!
- not tested any archiving methods !
- not tested with any tape technology upgrades preserving the
backups !
- backup to disk was tested only for TSM ! Why (try to guess)?
- etc., etc., etc. ...
3. Having the necessary hardware I am able to setup this four server test
environment:
- 4x 1 day - OS install;
- 4x 1 day - installation and setup of the product I perfectly know
(TSM for me and NetBackup for this unnamed Veritas engineer);
- 1 week to read the manuals, redbooks, etc. for the other product
(I have not ever heard about);
- 4x 3 day - installation of the "other" product;
- 4x 1/2 day x 2 (or 3) platform/product test.
This totals about a month of one-man-show. What about those two and half
months between 02.02 and 15.04 ? What took so much time? Should I include
in the timesheet more than a week product tuning for *EACH* platform
separately? Or the competitive product was so hard to understand that we've
spent month and half to read the manuals?
4. What is the purpose of this Altiris software quoted on the end of the
report? To quickly deploy somehow UNIX using tools targeted for PCs in a
classroom with/without specific tuning?
Zlatko Krastev
IT Consultant
"Malbrough, Demetrius" <[EMAIL PROTECTED]> on 21.12.2001 22:15:07
Please respond to "ADSM: Dist Stor Manager" <[EMAIL PROTECTED]>
To: [EMAIL PROTECTED]
cc:
Subject: Re: TDP for exchange vs Veritas hot backups
Also, Gerald! This may be off the topic but there was a nice analysis
done on TSM vs. Veritas! You can view it at:
http://www.keylabs.com/results/veritas/veritas.html
Regards,
Demetrius Malbrough
UNIX/TSM Administrator
-----Original Message-----
From: Gerald Wichmann [mailto:[EMAIL PROTECTED]]
Sent: Friday, December 21, 2001 1:44 PM
To: [EMAIL PROTECTED]
Subject: TDP for exchange vs Veritas hot backups
I find this is the hardest thing to sell among potential customers who
are considering both platforms (TSM vs Veritas). The reason tends to be
that Veritas's hotbackup solution will do a higher level of granularity
in restores - something companies in our experience definetly prefer.
Being able to restore a mailbox seems to weigh heavily in favor of
Veritas - or at least that's been my experience.
Now I know some of the arguments against it. E.g. it's not officially
supported my Microsoft and therefore if Microsoft makes a change to the
backup API or to Exchange, it could break their hot backups altogether.
This argument I find is of limited use because most companies who've
been using it have never experienced this problem and can easily argue
right back that Veritas will quickly patch the product such that it does
work.
One of the arguments regarding mailbox restores is also to use
Microsoft's deleted retention parameter so that the exchange server
holds onto deleted messages for a period of time before officially
purging them. That way they are recoverable without having to do a
restore of the server. My understanding of this feature though is to do
a mailbox recovery, it is still very painful and involves another
exchange server? I would love it if someone more knowledgable could give
me the low-down on how the deleted retention actually works when a mail
or mailbox needs to be recovered once it was deleted. How do you go
about recovering them?
Another argument is the speed of backups and restores. My understanding
is Veritas's hotback up much slower to do backups and restore. However
I've never seen any numbers nor have I had the opportunity to test this
myself. Exactly how big a difference is this? Does anyone have any real
tests or numbers or is this also a fairly week argument?
Thanks for the help!
