backup

chop · Fri Oct 21, 2011 5:15 pm

Is it safe to copy live files across the network i.e .ibv, .ibvm, .ibvd across the network using something like robocopy? or would I need to stop starwind. These are not live files in the sense of using reading/writing at the time of backup.

I hope that makes sense.

Cheers

Chris

chop · Fri Oct 21, 2011 6:01 pm

Got the answer myself, I need to stop then service and then copy the files. Will try and get a batch file sorted for this one.

Once the service is stopped, will it be safe to copy the content?

Fri Oct 21, 2011 6:28 pm

You should not do what you're doing - stopped service or not. StarWind has built-in replication. Use it.

chop · Fri Oct 21, 2011 6:34 pm

We are on the old license agreement of 2 mirror's and we never used it due to slow performance (This was our fault not the product), we do not have a ha license. Unless you can convert our current license agreement?

Fri Oct 21, 2011 6:38 pm

HA is HA and replication is replication. They are not the same and used for different purposes. We're going to deploy third-party IP WAN accelerator soon and represent own new replication engine so we may solve your peformance issues either way.

chop wrote:We are on the old license agreement of 2 mirror's and we never used it due to slow performance (This was our fault not the product), we do not have a ha license. Unless you can convert our current license agreement?

ed@zugzwang.com · Sat Oct 22, 2011 4:19 pm

Hi chop,
I'm curious to know if you were successful with stopping the service and copying the .ibv files.
Anton- you say we "shouldn't" do this", but why not? It sounds like a good plan for shops that can afford the downtime.
If the caches are flushed after 5 seconds, it "sounds" safe

Starwind has HA, mirroring, fallover, clustering, Snaphot/CDP, etc. But none of these is a simple backup of the files.
It's unclear (to me) what "replication" means here.
Ed.

btw- Anton, thanks for your patience and luciid explanations. Sometimes it may seem unappreciated, but we are fortunate that you pay so much attention to this forum!

Sat Oct 22, 2011 9:59 pm

Shops who can afford downtime are expected to run single StarWind node + remote replication built-in to StarWind. Stopping service does not guarantee any write caches are not purged for numerous reasons. It's "abort" way and not something we can even consider recommending using for near production environments.

That's no problem. I really enjoy talking to customers and treat this as a pleasant part of my job.

ed@zugzwang.com wrote:Hi chop,
I'm curious to know if you were successful with stopping the service and copying the .ibv files.
Anton- you say we "shouldn't" do this", but why not? It sounds like a good plan for shops that can afford the downtime.
If the caches are flushed after 5 seconds, it "sounds" safe

Starwind has HA, mirroring, fallover, clustering, Snaphot/CDP, etc. But none of these is a simple backup of the files.
It's unclear (to me) what "replication" means here.
Ed.

btw- Anton, thanks for your patience and luciid explanations. Sometimes it may seem unappreciated, but we are fortunate that you pay so much attention to this forum!

chop · Sun Oct 23, 2011 6:48 pm

Hi ed,

Gave up in the end, remote replication or cdp is not what I want. I simply want to make a full backup of an iscsi target. Replication helps, but is not the way I want to move the data. As this is not live data we use iscsi for backups only. So any content will be written overnight to the iscsi drive and this is what I want to grab and move across the network to a very large NAS.

But not to worry, suppose you cannot have everythign in life.

Sun Oct 23, 2011 8:13 pm

Snapshot replication is designed to solve exactly your task - copy accumulative amounts of data @ the background.

chop · Sun Oct 23, 2011 9:22 pm

That does sound great, but when creating a new snapshot/disk and you choose the journal location, you are only allowed to use local devices and not network based file storage. Or I am totally missing the point?

The snapshot/journal is no good as a backup solution if the data is held on the same device.

Cheers

Mon Oct 24, 2011 9:29 am

1) Right now you need to use Mirror to have data replicated to remote node.

2) You definitely need to have different physical media as a back end for primary storage and backup. It's not even discussed.

chop wrote:That does sound great, but when creating a new snapshot/disk and you choose the journal location, you are only allowed to use local devices and not network based file storage. Or I am totally missing the point?

The snapshot/journal is no good as a backup solution if the data is held on the same device.

Cheers

ed@zugzwang.com · Mon Oct 24, 2011 8:49 pm

Hi Chop,
I share your frustration. One of the benefis of software iSCSI should be that the data is in an .img file and and you can do what you want with it. Apparently you can't even make a copy of it. Even my VSS enabled backup software fails with the service stopped.

To my "Fat Ears" many of these terms sound like the same thing- HA, Snapshot/CDP, Mirror, Replication, etc. But of course they are not. ( I can discard HA, I know I don't need it.) I have been thinking of Replication as meaning a 1:1 duplicate copy of the .img file. So, to me, Mirror = Replication. ([Q1] Do these terms mean the same thing?) I thought I had it figured out until I read- "SNAPSHOT REPLICATION is designed to solve exactly your task - copy accumulative amounts of data @ the background." (my emphasis) I hate to sound so picky here, but these definitions are crucial to understanding the options.

In his recent post, Anton said- "The snapshot/journal is no good as a backup solution if the data is held on the same device". Ok yet, even if you set journaling to another device, it seems to still require the original file, so S/J can Never provide "backup" ([Q2] Is this correct?) Chop- I wonder if you can fool the "local disk" requirement by directing the journal device to an iSCSI drive on a different computer

)

I think Mirror = Replication. Anton says- "you need to use Mirror to have data replicated to remote node". This sounds like the real solution since the mirror can be an iSCSI on another computer.

I think what I need is called- Asynchronous Mirroring and Remote Replication (Active-Passive)
However in order to get it I need the Starwind HA version that supports- Synchronous Mirroring (Active-Active High Availability)
[Q3] Do I have this right?
Unless, of course, somebody can tell me how to simply make a copy of an .img file

thanks, Ed

Tue Oct 25, 2011 2:32 pm

Guys the task you're trying to solve with stopping the service and using third-party software to copy IMG can be solved by StarWind itself. Without stopping the service, terminating connections and using any extra tools.
It's called "Mirror". You need to configure "Mirror" plugin instead of using HA or plain single node IMG file. Upcoming version will allow to replicate not only IMG but snapshots thus being more effective. That's all. No frustration.