Advice on Cloning Needed

[bamber]

We currently run Legato 5.1 (looking to upgrade to 6 in the near future).

Before doing this I need to look at cloning, we run Legato on a Sun E250 with 135 clients predominantly Sun with the odd HP and NT clients included.

We use MTI raid in general (Vivant s200 1.7 TB) divided up into various size luns, the largest at the moment being 250GB.

My largest client takes over 24 hours to back up when performing a full backup.

Does anybody have any clues as to a formula or rule of thumb that I can use to work out how long the clone would take to run per Gigabyte etc.

Also is it easier/better to use Manual or Automatic cloning.

Many Thanks in Advance

Bamber

 

[jsargent]

I don't know of any formula, but it seems to me that the speed of your tape devices and the number of tape devices used will be the deciding factors. My tape library has 3 DLT 7000's, so I can only clone one volume at a time. As a guideline - our clone process takes about 2/3 of the time it takes for the backup to run. Note that these drives are all directly connected to the same backup server...if you're cloning over the network, obviously you wouldn't expect those results. Sorry I couldn't be more specific, but I hope this helps a little.

 

[ag100]

It depends.. If you let the group clone when finished, it will take longer, as you limit the number of streams that can run at once..

If you create a script to clone (match heads and tails, to reduce tape mounting and ejecting..) you can get through it a lot faster, assuming you have a bunch of tape drives..

Let me know if you need anymore info..

Good Luck!

-ag100

 

[BillAm]

I implemented several cloning scripts from The Networker Journal. These scripts took a bit of tweaking to work in my site, but they have REALLY helped our cloning process. Take a look at

http://www.drenpress.com/extreme_cloning.html

good luck!
-Bill

 

[eskolnik]

Can I get a copy of these cloning scripts. Denpress is no more.....

Ed

 

[Speridal]

look here:

http://www.drenpress.com/articles/extreme_cloning/extreme_cloning.html

 

[eskolnik]

I tryed that already but the "The Networker Journal is Closing
A serious medical condition with one of our founding partners has forced The Networker Journal to discontinue new publications.

We apologize to our faithful subscribers, but we simply cannot continue publishing The Networker Journal. The entire Dren Press staff has had a great time providing information on the Legato product line and we appreciate all of the positive responses we have received from our readers"

Again can I get a copy of those scripts?

Ed
Ed Skolnik
The Interpublic Group Of Companies, Inc.
GIS Chicago System Administrator
Chicago, Illinois 60611

 

[mcurran1]

I too would like to get the script info. Since they are not publishing could someone put these up?

 

[eskolnik]

I guess no ones talking. Ed Skolnik
The Interpublic Group Of Companies, Inc.
GIS Chicago System Administrator
Chicago, Illinois 60611

 

[Kinnison]

I was able to place an order successfully with Dren Press. I have not yet received the Volume 1 I ordered, however I do not expect it till next week. I will let you know when and if I receive it.

Paul Gilles
Navigation Technologies

 

[605]

O.K., most people do hard in understanding the NW cloning
process and i have to admit that NW itself also does not
help in making understanding easy.

First, NW can clone as fast as it has backed up. And this is
a serious and true statement. Just assume that you have a
media with one only save set: If you have to clone it, this
is the scenario you will end up with. The drive's speed is
the bottleneck. It can be the source or the destination device, whatever is slower. Please keep in mind that you
copy from a media to a pool, so the clone media might be of
another type than the source media you used for backups.

However, most backup media will consist of multiplexed
(interleaved) save streams and this is what the situation
makes more complex. Can you also clone such volume in a
multiplexed way (in one pass), creating a multiplexed tape?
Yes, in general. And this is Networker's default way how to
to cloning. However, this rarely happens due to other
parameters in the scenario. The general problem here is
called "tape drive contention" or "the fight for drives".
Such usually happens if you have overlapping save groups
and you are using automatic cloning.

-----------------------------------------------

On the other hand, in most cloning scenarios, you will not
clone the whole media, but only certain save sets. In this case, cloning will slow down because it will consolidate
the save sets (it will read the save sets sequentially while cloning). BTW, this is analog to a recovery process
where it is most unlikely that you will recover all
multiplexed save sets at the same time, especially not, if
they come from different clients. As sequential cloning
results in a decreased throughput (while de-multiplexing,
you must read through data that does not belong to this
certain save set) the cloning speed slows down and you need
more time.

Just think about an easy example:
- You are backing up 4 save sets of 10GB size equally
distributed. This results in 40GB of multiplexed data.
Let those be the only save sets on the backup media.
- If you do a volume clone, NW (if it can), will clone
the source media in one pass, creating a multiplexed
clone media.
- If you only clone 3 save sets, NW will most likely do
this sequentially and therefore will read through the
tape 3 times instead of one. So the cloning speed
reduces to 33%.
So there are a lot of things to consider if you investigate
cloning. And each scenario is different.

The most easiest way to prevent multiplexed data is to back
it up to a file device and clone the save sets from here. As
file devices never mix (interleave) save streams, such data
is ideal for cloning to tape (and for recovery, in general).

Hope this scheds some light.