Offsite Backup Solution Needed
-
In reading DenisKelley's post and then re-reading JB's post - I suppose you could go for incremental replication at the hypervisor level, but I don't know if that requires a server with a hypervisor on it at the remote location. Be that as it may, it's not a backup, it's a replication of the live system. So you can't go back in time. If the live system gets infected, and that infection is replicated to the DR site, you're done. So again, this is not backup.
How much data are we talking about?
You mentioned that your server crashed (I'm assuming your VM host) because it ran out of space due to snapshots?
Wow - what's your change rate? How much total data do you have?
I don't know how much extra storage you have on that server, but if you're running it out of space because a snap file is there, damn. Sounds like you have a huge change rate going on. Depending on your change rate, you might not even be able to replicate your backups over night with a 5/5, mathematically it might not work out.
-
@MattSpeller said:
In a similar situation with limited bandwidth like that we just used Symantec to dump to tape. Nice people in a van came by each week to get the tapes.
Which a great bon-fire is held weekly?
-
Yeah so there are a few different items being discussed.
Continuous replication is not a backup. It's an Oh-Shit recovery tool, where you are making a ready to boot copy of everything on a separate host.
- this does not sound like what you want
Backups include
- Full Backups - backing up everything VM related
- Incrementals or Delta's - Only the changes since the last backup.
Incrementals are what you appear to want, but then you mention that you'll have a Hypervisor at the remote location.
So are you doing / hoping for a Continuous Replication and Backup scenario where you use two types of recovery?
-
@MattSpeller said:
Nice people in a van came by each week to get the tapes.
How do you know they were nice? Did you actually speak to them?
-
@BRRABill said:
@MattSpeller said:
Nice people in a van came by each week to get the tapes.
How do you know they were nice? Did you actually speak to them?
Yeah they wouldn't take tape from anyone but IT staff, checked ID every time. Polite and professional.
-
@BRRABill said:
@MattSpeller said:
Nice people in a van came by each week to get the tapes.
How do you know they were nice? Did you actually speak to them?
10 years ago when I was using tape, we had Iron Mountain come by twice a week and swap boxes of tapes in a rotation. That guy was nice. Now I have a better system where I copy backups offsite to Amazon S3 and Glacier.
-
@Dashrender said:
In reading DenisKelley's post and then re-reading JB's post - I suppose you could go for incremental replication at the hypervisor level, but I don't know if that requires a server with a hypervisor on it at the remote location. Be that as it may, it's not a backup, it's a replication of the live system. So you can't go back in time. If the live system gets infected, and that infection is replicated to the DR site, you're done. So again, this is not backup.
How much data are we talking about?
You mentioned that your server crashed (I'm assuming your VM host) because it ran out of space due to snapshots?
Wow - what's your change rate? How much total data do you have?
I don't know how much extra storage you have on that server, but if you're running it out of space because a snap file is there, damn. Sounds like you have a huge change rate going on. Depending on your change rate, you might not even be able to replicate your backups over night with a 5/5, mathematically it might not work out.
A replication of a live system is completely fine, might actually be nice.
We have our onsite backups that go back 4-8weeks so infection isnt really our concern with the offsite, our concern is if we lose the building.
Total data is pretty big. If I was to guess...4TB? plus or minus 1TB (2TB being FileServer)
Change....as well probably pretty big, the biggest part right now (cause of how we do it but it can be changed) is DB dumbs of our SQL, so we would obviously change that, all servers change 100-200GB? maybe more, honestly not sure, DOESNT NEED TO BE EVERYDAY
Talking every week / every month (for less important)
Just plain and simple, whats the least painfull way to get backup if a plane flew into our building.
If our sharepoint site was 3 weeks old, but our POS was 1 days old, we're not doing too bad.
-
Did you ever talk to veeam about why snapshots were crashing your server? Do you have really old/under-powered hardware with super slow hard drives?
-
@wrx7m said:
Did you ever talk to veeam about why snapshots were crashing your server? Do you have really old/under-powered hardware with super slow hard drives?
Havent talked to them no, its was pretty black and white VM crashed due to resources, delete snapshot and poof it works.
(almost) everything is running off of quick systems, quick SAN's mainly all 15k or SSD
-
@Sparkum 100-200GB per day in changes?
That's a very large Delta if you're trying to replicate those changes off site.
-
@Sparkum I have been using Veeam since version 6.5 and now on version 9. I absolutely love it. I would still take a look at the reason you ran out of resources. It seems really odd that that you would have that problem on newer hardware.
I also have a large file server. About 2 TB is used and with version 9 and vSphere 6, the full backup only takes about 16.25 hours. On version 8 it took twice that long.
-
@DustinB3403 said:
@Sparkum 100-200GB per day in changes?
That's a very large Delta if you're trying to replicate those changes off site.
yeah do the math
5 Mb/s = 18,000 Mb/hr (2.25 GB/hr) max. It's unlikely that you'll get max use, assuming 80% you looking at being able to send 1.8 GB/hr. Assuming you close at 5 PM and open at 7 AM, that's 14 hours you can transfer at full speed, 1.8 * 14 = 31.5 GB per night.
-
@Dashrender That's also assuming all work halts at 5PM... no last minute changes....
-
@DustinB3403 said:
@Dashrender That's also assuming all work halts at 5PM... no last minute changes....
lol of course LOL
this was of course a best guess type scenario.
-
@Dashrender I really think it's tape or nothing with such a low bandwidth environment.
shrugs
-
So, the main problem here is your WAN connection's bandwidth. There is no chance you can get something better?
-
If you really have 100-200 GB worth of changes a day - there is no way you're replicating that over a 5/5 pipe, just not happening.
With that amount of change, I think you should consider tape and iron mountain. That will probably be your cheapest option.
If you can bump your internet to 50/5 on the server side, and 5/50 on the other side... maybe this would work, but man.. I wouldn't like that.
-
@Dashrender said:
If you really have 100-200 GB worth of changes a day - there is no way you're replicating that over a 5/5 pipe, just not happening.
With that amount of change, I think you should consider tape and iron mountain. That will probably be your cheapest option.
If you can bump your internet to 50/5 on the server side, and 5/50 on the other side... maybe this would work, but man.. I wouldn't like that.
Symmetrical or you're absolutely hosed if you actually need your backups
-
@MattSpeller said:
@Dashrender said:
If you really have 100-200 GB worth of changes a day - there is no way you're replicating that over a 5/5 pipe, just not happening.
With that amount of change, I think you should consider tape and iron mountain. That will probably be your cheapest option.
If you can bump your internet to 50/5 on the server side, and 5/50 on the other side... maybe this would work, but man.. I wouldn't like that.
Symmetrical or you're absolutely hosed if you actually need your backups
well it's across town - he already said he'd drive there, and that is why Carbonite was off the table.
-
@Dashrender ah good call, didn't see that.
