P2P backup solutions?

Associate
Joined
4 May 2011
Posts
1,065
One of the chief complaints of the users in our organisation is that there is no backup solution - there is actually, but its not automatic, so it doesn't count in our users eyes :) We have policies and processes in place to help users perform backups, but of course they don't pay any attention until after their hard drive has failed catastrophically.

The other related issue is that when a user gets a new machine, they have to transfer their own data to the new machine, because it isn't stored centrally, it doesn't just appear. This means tracking down USB sticks and all that fun stuff.

I've come across the concept of P2P storage and I'm going to investigate it, so I thought I'd check here to see if anyone had any experience with it. In my head, the solution looks like this:
-Software is deployed to each client through the software management system
-Each machine then automatically creates a "Backup" folder in the "My Documents" area
-Anything placed into this folder is encrypted and distributed to multiple peers
-Anything that's backed up that is not present on the system is automatically pulled down (So users getting new machines would pull down all their files automatically)

I realise that to have backups be remotely reliable, I will need to allocate 2/3 times more storage on each machine than I provide backup space for. I also realise that this is not bullet proof, that if all machines holding data die for some reason, that the data is lost.

To pre-empt a couple of questions:
I cannot use 3rd party cloud storage (Drop box etc), because security will not allow it. End of. Full stop. Conversation will not be entered into etc (Security hate 3rd party cloud)
I'm appreciate that there are other options and these are also being investigated, but for the purposes of this thread, I am only interested in the pros, cons and experiences of P2P backup.
 
I'm not sure it'd be worth the setup. Why have so many backups? Surely a backup to a server/NAS which then backs up to a disk (or a rotation of 5 disks for example) that's taken off-site every night will be just as secure?
 
Sorry, should have mentioned this. We have 8000ish users. If this works, there is potential for it to be adopted by the parent company, some 60,000 users. Its why we don't simply have a central automated backup - the costs would be astronomical, not to mention we would require dedicated data centres to manage it. The initial impression of p2p backup is that because it utilises existing but unused space, its basically free.
 
Interesting, so you'd have to find something that spoke to all the other clients, asked it if it had free space for the file and then copied it if so, or something along those lines.

Is the company spread over multiple offices? As long as they're all linked with a decent connection then that would sort the off-site backup issue (in case of fire or whatever). What if a particular file, for some reason, only gets backed up to the same office? You would need to make sure that doesn't happen else that file's gone in case of disaster.

Anything that's backed up that is not present on the system is automatically pulled down (So users getting new machines would pull down all their files automatically)
This could cause problems as well. What if a user deleted something? It'd keep coming back. You'd need to somehow tell the backup system to keep it but not re-copy it.

A quick Google brings up this: http://sourceforge.net/projects/p2pbackupsmile/. No idea if it's going to do what you want or if you'd even be allowed to use it (software licencing etc.), but might give you some ideas if you play around.

It's a shame they're not going to allow cloud backup. It'd be the perfect solution. Off-site, multiple copies and when the user gets a new machine the backup software is installed and files downloaded automatically.
 
Last edited:
Either that or a system that is managed by a central server, but yes, that's the idea.

Yes, we are over a number of offices/locations and generally have good connectivity. However, this isn't intended to be a disaster recovery option - those are going to cost serious money whatever you do and are outside the scope of what I'm looking at. This is only meant to cover the "Damn, my hard drive died, and I was just going to do a backup tomorrow I swear" crowd. If the software is configurable to prefer different sites based on subnet or similar then great, but its not a primary concern.

And yes, I acknowledge that this is going to involve educating users not to use this to store business critical data. The reality is that we have some users who do backup to SharePoint like they should, who will pay attention, and we have some users who ignore anything IT related that they don't NEED to do in order to do their jobs. Best we can do for these users with this solution is provide the backup and hope that anything critical will be recoverable.

While I agree a cloud solution seems ideal, its also very expensive. Take for example dropbox. Their business pricing is 125USD per user. That would work out to 7.5million USD if adopted across the company.
 
Not to sound too harsh (Warning: this is definitely going to sound harsh, but please don't take it personally), your company IT needs to grow a pair and put a proper policy and controls in place.

You're trying to use more technology to solve a problem that you already have the technology to fix. Lockdown PC's, force folder redirection, and create shares or use sharepoint for users to save their files to.
At least then you only have a few servers to ensure backups for, rather than 8000+ users PC's, tablets and who knows what else.

If you're going to have to re-educate users to save stuff to be backed up to a particular folder, better to have them save to a network share and have done with it.
############
Once the red mist had cleared and I reread your post I realised you said you do have a policy and a process in place. The problem is you have no controls, and that's what you should be addressing and putting in place. By doing this local PC cloud backup stuff you are actively working against your own policy (assuming it says no data is to be saved locally).

Also, this sounds like it's not a backup, it's a sync. Wait till a user deletes a file accidentally, or saves over a document and doesn't notice till after a sync - you have no way of recovering it. If this happens to the CEO, you might be looking a for a new job if this is touted as a 'backup' system.

Again, I'm sorry if this has come across as rude, but I have been in the same position (Only 3,000 users though), with no PC's locked down, everyone with admin rights and can save anywhere, so I'm sympathetic to your plight. It took a massive incident to force management to realise the need for change (despite who knows how many business cases, etc being ignored). It took 18 months to achieve, but it has been seriously worth it.
 
Last edited:
Little_Crow - Don't worry, I know its not personal. Much of what you say is true. However, as I mentioned, this is just one workstream among many being investigated as a solution. I don't envisage this being the final answer to the problem, but rather a potential piece of the puzzle.

For example, lets say we lock down all the desktops so that remote saving is the only option. Our existing SharePoint infrastructure can't support this, so do we upgrade all our server infrastructure at significant cost? Provide share drive capacity equal to 8000 x hard drive size at significant cost? Provide a private p2p 'sync' (Good point on that one :)) at ??? cost?

The point I'm making is that I know this isn't the solution on its own, which is why I want to know what it can and can't do, so I can feed it back into the 'bigger picture™'
 
Linkex said:
We have policies and processes in place to help users perform backups, but of course they don't pay any attention until after their hard drive has failed catastrophically.
User problem. Wave the policy in their face and tell them to go away. I know it's not helpful but you're going to be placing an enormous administrative burden on the company otherwise.

Linkex said:
The other related issue is that when a user gets a new machine, they have to transfer their own data to the new machine, because it isn't stored centrally, it doesn't just appear. This means tracking down USB sticks and all that fun stuff.
Do they not have a home drive with a bit of storage? If not, get them and enforce storage quotas. 500MB x 8000 is only 4TB of storage which is not much at all. That's enough for most people's Word documents and PDFs. Hell, even 10GB is only 80TB, not that expensive for an organisation as large as yours. People don't need that much storage on the system, and if they do, you should be providing it for them - they need the right tools for the job. Make it clear that they shouldn't be saving personal data on the system.

You're trying to decentralise a problem which is better solved by being centralised. As Little_Crow rightly says, how do you recover stuff in a speedy fashion with that system? Find whatever random node the data ended up? You're reinventing the wheel. Previous Versions in Windows does exactly this.

Sharepoint can support it, it's just a case of throwing more BLOB storage at it. Whether or not you want the data to be funneled through your Sharepoint system is another matter (personally, I wouldn't).
 
We've around 250,000 users spread across dozens of sites. All users information is remote and nothing is saved on the local machines except for a few cookies etc (which are also on the remote profile).

Comes a time when a new process is brought in, or new software (for example, the move from folders to sharepoint). A line in the sand is drawn for users to migrate ALL their data to the new system, anything not migrated will be deleted.

If the company wants this to work properely, it's going to have to invest and get it done. At least if you have storage centrally (centrally to each site perhaps) then it makes a lot of things easier such as administration and backups and will probably save money in the long term (I know nobody looks longterm nowadays).
 
The storage you need to centralise this is going to basically cost nothing. A 40TB (Usable) NetApp 2240-4 is under 40k. You could do similar for less (and I doubt you need 40TB per location anyway), but snapshots and SnapMirror is going to provide you some pretty painless data recovery options.

The storage in those desktops isn't free anyway and aside from all the good things posted above, this P2P approach will cause you issues if/when you want SSD in your client PCs unless the volume of data backup is particularly small.
 
Don't let your users force you into a solution that will multiply support costs exponentially.

Data shouldn't be on local drives (ever, in my opinion) it's not secure and a nightmare to control and backup.

Centralise data and redirect users to this. Disable direct user access to the local drive. Once all your data is in just a few places (or preferably only one place!), backups and restores become a tiny part of administration and while your users won't like it in the short term, the long term benefits when they find out they can get files back easily will outweigh this.

Heck if you choose the right file server platform the users can self-serve from snapshots using the built-in "Previous Versions" tab in Windows. We have this in place (snapshots taken up to 4 times a day) and since implementation support calls for file restores have reduced by 90%.
 
Last edited:
Don't let your users force you into a solution that will multiply support costs exponentially.

Data shouldn't be on local drives (ever, in my opinion) it's not secure and a nightmare to control and backup.

Centralise data and redirect users to this. Disable direct user access to the local drive. Once all your data is in just a few places (or preferably only one place!), backups and restores become a tiny part of administration and while your users won't like it in the short term, the long term benefits when they find out they can get files back easily will outweigh this.

Heck if you choose the right file server platform the users can self-serve from snapshots using the built-in "Previous Versions" tab in Windows. We have this in place (snapshots taken up to 4 times a day) and since implementation support calls for file restores have reduced by 90%.

I'm really wary of giving users the ability to restore their own files (the average user, that is), but it does certainly have its advantages!
 
My Head of IT took some convincing, but problems have been minimal. Snapshots are read-only so old copies are always available even if they overwrite the wrong thing. We don't allow restores at folder root level just to be safe :)

In any case, you have to actively turn it off on Windows 7 so it's less work to just allow it. On the cost side, it's less expensive to have a server admin go fix something if it goes wrong than it is for EVERY file restore to require a server admin.
 
Back
Top Bottom