Backup and Recovery Goals
-
Obviously Porsche would be the Gold option, and Ferrari would be the Platinum option.
-
You have a bit of old equipment that should be able to be reused here.
Again, back to my suggestion:
1 new server for all production workloads1 local onsite backup server (running StorageCraft) - maybe a new to you/used box if current hardware won't work
1 offsite at your remote location replicated with the StorageCraft box - use currently owned hardware
-
@DustinB3403 said:
So I found a rather easy way to find our delta for last week. 12.52 GB. Which isn't that much. Pushed out hourly to an online storage service could be reasonable.
Now to find options..
12.52 GB = 100.16 Gb = 100,160 Mb Assuming the 30 Mb/s connection mentioned earlier (and that you can use the whole thing) = 3338.7 seconds = 55.64 Mins
OK not bad, takes about 1 hour a day at full saturation to push that to your remote site. As you said, if you do that hourly, you should be fine.
-
Alright @scottalanmiller, Dustin seems to really want his full backups. Am I crazy to think they really aren't needed considering he has a StorageCraft backup solution?
I know this is a point of difference of opinion between Dustin and I.
-
@Dashrender said:
@DustinB3403 said:
So I found a rather easy way to find our delta for last week. 12.52 GB. Which isn't that much. Pushed out hourly to an online storage service could be reasonable.
Now to find options..
12.52 GB = 100.16 Gb = 100,160 Mb Assuming the 30 Mb/s connection mentioned earlier (and that you can use the whole thing) = 3338.7 seconds = 55.64 Mins
OK not bad, takes about 1 hour a day at full saturation to push that to your remote site. As you said, if you do that hourly, you should be fine.
Which can be done at night. Kick it off two hours after people have gone home and you are good to go.
-
@Dashrender said:
Alright @scottalanmiller, Dustin seems to really want his full backups. Am I crazy to think they really aren't needed considering he has a StorageCraft backup solution?
I know this is a point of difference of opinion between Dustin and I.
Weekly fulls? No, not needed.
-
@Dashrender and @scottalanmiller how should I backup my VM's then? What is the restore process for if a VM takes a dive and I have no full to recover from?
The backup process has to remain someone current.
-
@DustinB3403 said:
@Dashrender and @scottalanmiller how should I backup my VM's then? What is the restore process for if a VM takes a dive and I have no full to recover from?
The backup process has to remain someone current.
StorageCraft creates synthetic full backups from the incrementals. Veeam, Backup Exec, Unitrends, etc all do this.
You mount the date/time of the restore you want to perform, and say - do the whole enchilada.
-
I'm not familiar with StorageCraft specifically, so I speak about AppAssure.
With AA I boot from a CD in the VM I want to restore to. That image finds my AA server on the network, I authenticate then choose my restore point, and tell it to restore the whole thing.
-
@scottalanmiller said:
@Dashrender said:
@DustinB3403 said:
So I found a rather easy way to find our delta for last week. 12.52 GB. Which isn't that much. Pushed out hourly to an online storage service could be reasonable.
Now to find options..
12.52 GB = 100.16 Gb = 100,160 Mb Assuming the 30 Mb/s connection mentioned earlier (and that you can use the whole thing) = 3338.7 seconds = 55.64 Mins
OK not bad, takes about 1 hour a day at full saturation to push that to your remote site. As you said, if you do that hourly, you should be fine.
Which can be done at night. Kick it off two hours after people have gone home and you are good to go.
It could be, but if he's extra paranoid, or just wants to make sure the sync is more than once a day, he can kick it off hourly.
Assuming the typical work day (for production of files) you're dividing that hour into 9 chunks, so it should take around 7 mins at full speed, probably more like 12 at not full speed.
-
@DustinB3403 said:
@Dashrender and @scottalanmiller how should I backup my VM's then? What is the restore process for if a VM takes a dive and I have no full to recover from?
The backup process has to remain someone current.
You have a full ready to go that is older, say monthly, and you apply the delta to it. Instead of incrementals, move to differentials. All you need is the full plus the differential. Inrementals mean you that you need EVERY incremental to do a restore. Differentials mean you only need the Full plus the latest Diff.
-
@scottalanmiller said:
@DustinB3403 said:
@Dashrender and @scottalanmiller how should I backup my VM's then? What is the restore process for if a VM takes a dive and I have no full to recover from?
The backup process has to remain someone current.
You have a full ready to go that is older, say monthly, and you apply the delta to it. Instead of incrementals, move to differentials. All you need is the full plus the differential. Inrementals mean you that you need EVERY incremental to do a restore. Differentials mean you only need the Full plus the latest Diff.
But you didn't mention that StorageCraft is what has that full. I say this just to be clear.
@scottalanmiller Does Unitrends ever take a new full backup as a matter of standard practice? (I'm not considering issues where there might be corruption or something that causes you to need another full backup)
-
@Dashrender said:
@scottalanmiller Does Unitrends ever take a new full backup as a matter of standard practice? (I'm not considering issues where there might be corruption or something that causes you to need another full backup)
Yes, regularly.
-
@scottalanmiller said:
@DustinB3403 said:
@Dashrender and @scottalanmiller how should I backup my VM's then? What is the restore process for if a VM takes a dive and I have no full to recover from?
The backup process has to remain someone current.
You have a full ready to go that is older, say monthly, and you apply the delta to it. Instead of incrementals, move to differentials. All you need is the full plus the differential. Inrementals mean you that you need EVERY incremental to do a restore. Differentials mean you only need the Full plus the latest Diff.
I can still run this full monthly with a tool like NAUBackup and a Crontab job. That is where I'm considering the fulls from coming from.
The incremental backups would be built from Storage Craft and for file level restore.
-
@DustinB3403 said:
@scottalanmiller said:
@DustinB3403 said:
@Dashrender and @scottalanmiller how should I backup my VM's then? What is the restore process for if a VM takes a dive and I have no full to recover from?
The backup process has to remain someone current.
You have a full ready to go that is older, say monthly, and you apply the delta to it. Instead of incrementals, move to differentials. All you need is the full plus the differential. Inrementals mean you that you need EVERY incremental to do a restore. Differentials mean you only need the Full plus the latest Diff.
I can still run this full monthly with a tool like NAUBackup and a Crontab job. That is where I'm considering the fulls from coming from.
The incremental backups would be built from Storage Craft and for file level restore.
But does breaking this appart from StorageCraft actually make sense? Assuming you're a Unitrends shop you typically wouldn't do that. You'd have Unitrends do everything.
-
Can Unitrends build a complete backup that can be imported to a Xenserver should the Guest VM get corrected?
-
And does it have to run on a separate server to create that backup file?
-
@DustinB3403 said:
Can Unitrends build a complete backup that can be imported to a Xenserver should the Guest VM get corrected?
Not sure what you mean. Unitrends handles every aspect of XenServer.
-
@DustinB3403 said:
And does it have to run on a separate server to create that backup file?
It would run as a VM. VMs have no means of knowing if they are local or not.
-
@Dashrender said:
But does breaking this appart from StorageCraft actually make sense? Assuming you're a Unitrends shop you typically wouldn't do that. You'd have Unitrends do everything.
i agree. Find one tool that does the job and stick with it. Trying to use two separate tools is making this incredibly complex and complexity is just another term for risky.
