|
OK, Al. I said I'm willing to learn and you said you're
willing to teach. So let me ask you what you would do in my place. I
ask it in general terms and hypothetically--I'm not trying to get a freebie
consulting engagement.
So you have 80 workstations, 100 users, 8 servers, Win,
Exchange, SQL server. That's the assets. The constraints: you
can spend maybe $5000 a year on purely backup solutions. If you really
raise a hue and cry, you might be able to get 3-4 times that.
[1]
But the budgeting of money is not the show-stopper. The
budgeting of time is the real killer. Figure on having less than half the
time you need, and that includes learning/training. I'm being generous
here.
That's the rules I have to play by. No
fair changing the rules. ;o)
BTW, I know nothing about SAN, except by rumor. I was
recklessly speaking above my pay grade.
BTW2, It never occured to me to question your view of the real
world. Your comments and analyses have alway struck me as extremely sober
and to the point. It's just that we are at different levels of the real
world. (how the heck does Susan do it, anyway?)
[1] Keeping in mind that security is putting increasing strain
on the 'protection' resources, and that management and users are far more
concerned with spam and viruses, which they can see, than with backup, which
they hardly ever see.
----- Original Message -----
Sent: Friday, November 03, 2006 7:29
AM
Subject: Re: [ActiveDir] OT - Backup
Follies (was) Exchange Log files --Disk Full--
Yep, I'd have to disagree, but then again, I have a different
perspective. San based connectivity to the files is not something that
*should* be a problem for you. I understand how it could be, but from my
perspective that's a process issue. More specifically, that's a process
issue that is easily solved by ensuring that you have good controls on your
server build process.
Now before you say I don't live in the real
world because control over build configurations is something that's only a
dream, let me point out that I've been doing this for a long time and I'm well
aware of all kinds of processes and thoughts that may or may not be painful.
One such process is not having good solid control on the server configurations
that you accept into your datacenter/closet. In the case of SAN
technology, the use of HBA's makes it really easy to not care about the
drivers. But more to the point would be having the controls in place to ensure
that everything needed to rebuild/replace that platform is readily
available. If for no other reason than for disaster recovery
purposes.
If that's not the real world for you, I understand
that you're not alone. I also understand the consequences of that
behavior all too well and detest it. I've seen some of the largest
enterprises in the world employ similar "drop and pray" practices for that
behavior. Shops like Susan's likely have it much better.
Don't
let drivers and such get in your way to the appropriate solution for your
enterprise is all I'm saying.
On 11/2/06, Albert
Duro <[EMAIL PROTECTED]>
wrote:
Yes, BE does do disk backup. But I have some
objections:
A. They don't make it easy, infact they make an
unnecessarily complicated production of it.
B. I started doing NTBackup to disk while (and
because) I was still troubleshooting BE. When I gave up on BE and its
brethren, NTBackup was a natural segway, and already in place and
working.
C. I discovered one great advantage that
NTBackup-to-disk has over any other backup system: with a bit of
planning, it is proof against almost any combination of crash and
burn.
You have a backup file on two or more
disks/machines. Things go bad, you can do recovery from any Windows
machine; you can move or copy the backup disks/files to any machine.
Try doing that with a sophisticated tape-based or SAN-based system.
Imagine having to replace the tape drive/autoloader with the exact same
type, while rebuilding a same-hardware three-year
old server to the exact same configuration, same SPs, same backup
software, same drivers. I can guarantee that at least one of those
necessary replacement elements will be impossible to find, even under
leisurely conditions. [1] Yes, there are strategies to deal with
that, but if you could spend that kind of money, you would have gotten a
double-redundant bullet-proof system in the first place.
I truly hope that I'm wrong out of lack of knowledge and
pessimism. I am open to being corrected and encouraged.
[1] Naturally, the tape drive drivers will be on the same
tape that you can't access nohow. Download the drivers from the OEM,
you say? Chances are excellent that the OEM has gone out of business,
or sold out to a giant who prunes out what they don't like (and what
you need), or changed the name or version number on it out of sheer
orneryness. If you do get to what looks like the right drivers,
you're likely to find that the last minor upgrade version that really worked
well for you has been dropped, or tweaked into your trouble zone. I
can testify to ALL these experiences. I think others can
too.
----- Original Message -----
Sent:
Thursday, November 02, 2006 3:27 PM
Subject:
Re: [ActiveDir] OT - Backup Follies (was) Exchange Log files --Disk
Full--
Trying to remember exactly, but doesn't BE have an option
to use disk vs. tape drives?
You *could* run a test to help
simplify and rule out some of the complexity. Could take a while, but
might be worth it.
Al
On 11/2/06, Albert
Duro <[EMAIL PROTECTED]> wrote:
Why does NTBackup work for me and BE not, when they
are at core the same product?
I wondered about that too. Here are my
thoughts.
First, NTBackup is a simpler product that doesn't get
tangled up with the complexities of scheduling and a GUI.
But the real reason, I think, is that I've been
doing NTBackups to disk, while BE was to tape.
I've always suspected that most, if not all of my
difficulties with BE had to do with the drivers for the tape drives and
autoloaders, and with the SCSI interface to other devices ('other' being
anything beyond the normal HD and CD complement)
-----
Original Message -----
Sent:
Wednesday, November 01, 2006 6:11 AM
Subject:
Re: [ActiveDir] Exchange Log files --Disk Full--
Well put Albert. Thanks for that feedback.
What still has me curious is why BE wouldn't work in your
environment and why ntbackup does (partially at least). ntbackup
as written by the same exact people and has a lot of the same code
(it's licensed by Microsoft from Seagate last I checked). Ntbackup is
the less featured version designed for single host backups and
extended to act like it does more.
So that said, I agree that
the goal is that your client's data is backed up. I have to say
that I disagree that jury-rigs, mickey mouse and by the seat of your
pants is the long term solution though. That's an infrastructure
component that will come back to haunt at some point down the
road. As an interim fix, of course it can work. I'm not
blinded by the big vendors to the point that I think they have the
only solution. Far from it. But I like to think that I can
at least share some perspective and experience related to where it
leads and I definitely favor technology over layer8 processes.
Why? Because layer8 changes and grows out of current positions
and foundational solutions should not have to be decimated when that
happens. I've seen that way too often to care to see it continue
where possible.
Basically, I hate to see a foundational
solution such as backup, rely on such complexity and human
intervention. I completely understand that you have to do what
you have to do. When you wrote it in your original email, it sounded
like you approved of that method. Reading this last one, I can
you don't. I was just trying to point out where that leads and
trying to understand how you go there. I bet I would have gotten
there the same way you did ;)
Best of luck getting that worked
out.
If you need anything from me, please don't hesitate. I
have been known to make some backup solutions work :) Feel free
to ping off-line if I can be of any help.
On 10/31/06, Albert Duro <[EMAIL PROTECTED]> wrote:
Al, since you ask, no I don't see it differently,
at least not at the oratorical level. But where the rubber
meets the road, things can look very different. Like the military
say, the best laid plan falls apart the moment it meets the
enemy. You assume that I monkey around with Ntbackup and balky
media for economic reasons. In fact, we spared no expense
(relative to our small size) to put in industrial-strength backup
systems, both software and hardware. Even paid consultants to
set it up and manage it.
It blew up in our faces. Primarily because
Backup Exec just wouldn't work right in our environment. (I'm
not saying that BE isn't a fine product, it would just never work
for us). Why not? Don't know -- I couldn't figure it
out. Our consultants couldn't figure it out. Veritas
support couldn't either, nor the autoloader manufacturer. For
more than two years, nobody could figure it out, until I decided to
stop throwing good money after bad.
Did I try alternative products? In the same
class, yes -- more tales of woe, but different reasons. We did
not nor are we going to buy the high-end systems, which cost more
than our whole network is worth.
So I was left with NTBackup,
and admittedly a little more gun-shyness about brand-name
backup products than is strictly rational. That's what I
have to work with, and I try to make the best of it. That's
the 'real world' in my little corner of it.
Believe me, when you and joe and others on this
list urge us to 'make the best', I listen, I learn, and I
applaud. And it does push me in that direction. But the
only path there goes through 'make the best of what you've
got'. It's bumpy and often barricaded.
But after all is
said and done, the REAL point is that I am preserving my clients'
data and keeping them happy. Jury-rigs, mickey mouse, and
by-the-pants not withstanding.
-- Original Message -----
Sent:
Sunday, October 29, 2006 4:30 AM
Subject:
Re: [ActiveDir] Exchange Log files --Disk Full--
sub-optimal media are part of the real world?
Wow, thanks :)
Truth be told, that's a rant of mine. I've heard a lot
(lately especially) about how we want to do things cheap and
inexpensive and we'll fix it later and so on. I've also spent a
great deal of time cleaning up that kind of stuff.
Unfortunately, once it escapes into the "real world" then it
becomes more difficult to clean up because you have to do so in
front of customers/clients.
Interesting approach though. Usually a less disciplined from
what I've seen and often results in more expense related to
downtime and troubleshooting and lack of service. I'm
interested if you see differently though.
This area of the business fascinates
me....
On 10/28/06, Albert Duro <[EMAIL PROTECTED] > wrote:
I'm sure you and Susan are right. All
I'm saying is that it *can* happen, and for me, why take the
chance when one-job/one-task is easy to do.
Good point about the media, and that may
explain my case, but, hey, sub-optimal media situations are part
of the real world.
-----
Original Message -----
Sent:
Saturday, October 28, 2006 6:33 AM
Subject:
Re: [ActiveDir] Exchange Log files --Disk Full--
I've not had that same experience. Granted, it's a
limited feature utility (note the use of the word utility vs.
tool as requested) but it's still capable of doing more.
There were some fixes to ntbackup in service packs and
such. You might want to verify you're using the latest
version of that's what you see.
Also, check the media it's headed to. It's error
handling is not very elegant, but I've found it to be useful
and strong enough to stand up to some complex tasks in the
past. I've got several running now via cli that have been
in place for more than half a year without issue (I know, I
know, spend all that money on an enterprise backup system only
to backup some machines locally. But there are times
when it makes more sense, trust me.)
-ajm
On 10/27/06, Albert Duro <[EMAIL PROTECTED] > wrote:
I've found, with NTbackup, that if you
cram two or more tasks into a backup job, it's very likely
to fail. For example, if you do a System State and a file
backup and an Exchange backup in the same job. It's
best to separate each task into its own job, and sort it out
in the scheduling.
A mixed job will also work for a while and
then fail, which sounds like what happened to
OP.
-----
Original Message -----
Sent:
Thursday, October 26, 2006 2:21 PM
Subject:
RE: [ActiveDir] Exchange Log files --Disk Full--
Do you have multiple information
stores on this storage group? (If using
Exchange Enterprise edition)...the logs can't flush until
all stores have a full backup, because the logs are
shared...
--James
Hi,
I am running Normal
Backup. Using NTBackup Utility. Backing up Information
store.
Are you running full (AKA normal)
backups every night? It seems not. Use NTBackup to backup
to disk (obviously, you'll need a disk with over 120GB of
available space) and then use whatever normal program you
use to back that backup onto tape. This will keep you
running until you sort out why your normal backup software
isn't flushing the logs when the backup completes.
How are you currently running
backups? What software is in use? Are you sure it's
Exchange aware? Are you doing brick level backups or copy
backups instead of a full backup? Neither will flush the
logs.
I'd resolve this as quickly as
possible, because if you are in a situation where you have
to replay the logs, you're NOT going to be a happy camper.
Hi
All,
Kindly suggest, what i can do
about my Exchange Log files?
I have about 120 GB Log files
for past 4 months. I have a few doubts:-
Do i really need all those
log files?
If yes, Then how is it
possible to manage with this as i have a very limited
space left.
Can i delete these log
files?
Backup doesnt remove these
log files?
i am really running out of
space on my Exchange log storage drive.
Thanks!!!
Ravi
|
