To add to that if you use filelevel, I'm guessing you're referring to
file level backups without using vcb, you backup each individual vm on
the esx server. All the backup traffic goes through the one or more
network cards in the server. This can cause a bottleneck at the network
level on the server with users accessing the vm's through the nic and
the backup going through the nic.
If you use vcb, even if you use filelevel through vcb using "ALLVMFS" in
the saveset it should go through the fiber connections. That's if your
proxy server is also a storage node that's fiber connected to your tape
drive device. Plus with 7.6 you can do an image level backup and do
either an image level restore or a file level restore from it (windows
only). Plus it's faster in most cases to backup through fiber than it
is to backup over lan.
I agree with James that zoning your vcb server to be able to see all of
your esx luns is scary, but in the long run it's worth it. Just make
sure you create your drive that the vm data will be held on before you
zone the proxy server to be able to see all of the esx luns.
Here's my ocd kicking in here ---> We also install the networker client
on each vm, just in case the worst possible scenario happens and we have
to backup each individual client. There have been times where we've had
an issue with a particular vcb backup and it's taken a while to fix (a
while being weeks). During that time we still get a backup of the
client because we use a file level non-vcb backup for it.
-----Original Message-----
From: EMC NetWorker discussion [mailto:NETWORKER AT LISTSERV.TEMPLE DOT EDU] On
Behalf Of James Pratt
Sent: Tuesday, April 06, 2010 7:28 AM
To: NETWORKER AT LISTSERV.TEMPLE DOT EDU
Subject: Re: [Networker] Networker VCB vs filelevel
>> -----Original Message-----
>> From: EMC NetWorker discussion [mailto:NETWORKER AT LISTSERV.TEMPLE DOT EDU]
>> On Behalf Of kel AT STERIA DOT DK
>> Sent: Tuesday, April 06, 2010 4:14 AM
>> To: NETWORKER AT LISTSERV.TEMPLE DOT EDU
>> Subject: [Networker] Networker VCB vs filelevel
>>
>> Hi there
>>
>> Being alone with backups, I could use some sparring on the subject of
>> image backups with networker or just normal filelevels.
>>
>> Seems to me the VCB is redundant, save for the offload part, if you
do a
>> normal file backup includning VSS, you could just deploy a new
virtual
>> machine and restore everything back into that, instead of doing the
image
>> backup and then do the file restore.
>> Applications like SQL exchange notes etc, will still be restored with
the
>> networker user module..
>>
>> But am I missing something here ?
>> We have done serveral full restores, with the normal networker user
>> interface and it works fine, we rolled back a service pack install,
so as
>> far as I can see its a proven way to do it.
>> The VCB method, the new one in 7.6 is not stable yet, and while doing
a
>> restore of a full machine it crashed. So is VCB really worth the
extra
>> setup complexity ?
>>
>> I realise the offload aspect is worth considering too, but is it
really
>> that bad ?
>>
>> Any thoughts or experiences on this ? Am I missing something that
makes
>> VCBs worth the effort ?
Well here are my two cents, having done this for a few years now ....
VCB is by no means perfect (Zoning a windows box into your VMFS storage
group/luns is really pretty scary - and risky, so I've never liked that
aspect!), and is currently on the way out the door, to be replaced with
vsphere "Vstorage APIs" in Networker and other BUR products after
7/2010, but that's another topic...
It really depends on your site/environment - for example, we have many
vms that are not "Standard", meaning the C:\ drives may not all be the
same size , and of course many other things that make the large majority
of our systems "unique".
In our case, it behooves us to utilize VCB for the following reasons:
Faster recovery in the case we do not have VM templates ready to spin up
for "BMR" / ALL recovery like you speak of - for example, our print
server bit the dust (Spooler wouldn't start, bad drives) in the middle
of the day last week - users and then the help desk freak out of course
- boom, I crash old server, export-vcb files out of nw, re-import using
VMware converter or the native (Yeah, not so hot yet in 7.6 or 7.5
built-in legato VCB restore function) back into vcenter - power it up -
BOOM - the old print server is back online, literally within a half hour
and no one even knows the difference ... So, in our case, it would have
taken literally a few hours to get a new vm built out - (We also store
no templates for each vm due to storage constraints) before the "BMR"
recovery (Yes like you say it does work fine, but there is definitely
more involved, most importantly to us - the time to get it prepped prior
to restore is not worth it to us).
Another thing I like about VCB is that I don't even need to care or
worry about what version of windows it was I need to recover - R2, 2003
R2, 2003 SP2, etc) - as I said, boom, our print server was back online
in a half hour from the prior nights backup, and I still don't know or
care what specific version of windows it is/was...
In our case, we run 99% virtual. We *really* don't want to pay for NW
client licenses per vm, so we purchased the Virtual ESX CAL's for every
ESX host in our cluster. Boom, instant savings on licensing going
further, and no worries on future growth, outside of normal storage
considerations/time-windows and the usual suspects affecting the whole
BUR environment in general.
I could probably come up with other reasons , as well as more
disadvantages too - And yes, you will want to use "All" and the modules
for your "important servers", such as Exch, SQL, AD, etc etc...
Cheers,
Jamie
To sign off this list, send email to listserv AT listserv.temple DOT edu and
type "signoff networker" in the body of the email. Please write to
networker-request AT listserv.temple DOT edu if you have any problems with this
list. You can access the archives at
http://listserv.temple.edu/archives/networker.html or
via RSS at http://listserv.temple.edu/cgi-bin/wa?RSS&L=NETWORKER
To sign off this list, send email to listserv AT listserv.temple DOT edu and
type "signoff networker" in the body of the email. Please write to
networker-request AT listserv.temple DOT edu if you have any problems with this
list. You can access the archives at
http://listserv.temple.edu/archives/networker.html or
via RSS at http://listserv.temple.edu/cgi-bin/wa?RSS&L=NETWORKER
|
