JaguarPC bad backup restore practice

[pixelmac]

Hello,

My VPS is down for more then 24 hours now.
There was some sort of corruption on my vps.
bad things happen to any one in the industry, so at least we have backups.

JaguarPC use R1soft to backup their clients.
When you'll need to restore from a bad situation your VPS, you'll find out that it would take a day or two.

This is not acceptable by any standard.
If they had a local backup of the vps, they could just move it to a new one in no time.

All my clients are down, all very angry, two will leave me, and one sent me her lawyer.
JAG team must understand the importance of recovery time, and notify clients that R1soft backup takes a few days, and also recommend additional local backup option.

I'll appreciate if other JaguarPC clients will tell me how they handle their client's backups.

Thanks

Dan

[JPC-Awais]

Dan,

Your server has been restored and all sites are loading now. Your ticket has been updated with all details.

Restore time using r1soft is dependent on multiple factors, like size of the vps, or the other backups and restore tasks in progress at that time for same node. If from local backups you are referring to simple rsync backups then you are only looking at one aspect of restore time, those backups are no more practical at such a level, they are good for individual clients for taking their local backups. R1soft backups can restore in range of situations, ranging from a single file restore to a complete bare-metal restore of the server. Though yes we are working and exploring on all possible ways to improve the restore time for the data.

We always recommend clients to take local backups on top of the backups we maintain for emergency purposes. I would suggest you taking a look at "Back-ups and Data Loss" section in our TOS

[pixelmac]

Well,

All this is very nice technical info, but for real world situations, no one care about how it is done, as long as it's done fast.
My clients and I had to wait 25 hours for the restore to finish.
That is not acceptable by any standard.

Clients and costumers where on hold, they just couldn't work.
They don't care about R1soft, and how smart the service is.

You must find a better, and faster way to restore clients.

Thanks for the hard work you did by the way, I'm happy to be back online again.

Now I'm looking for a cheap FTP server, where I can let WHM send the backups automatically.

Best

Dan

[Frank Broughton]

Sure and we must find a way to live on Mars. Some things just take time. Spread out your clients on several servers. Have you own backup system and move clients around yourself in times of a down server.

No one likes a down server JPC probably more than any client they have, but it a fact of life and it "ain't gonna change" anytime soon.

Very frustrating for you I know - I have been through it too, not trying to jump on you - only to speak as to what is reality.

[jason]

Unfortunately, as hard drives have grown bigger (along with the amounts and type of data we store on them) the time necessary to do full restores from backups has also grown exponentially. Twenty-four+ hour restores may seem excessive, but when you take into consideration how massive servers have become in terms of processing power over the past few years and you add up the numbers, they grow quite fast. Add to that the fact that network speeds haven't really kept up with processing power. I don't know what JPC uses within their datacenter, but gigabit Ethernet has only been widely available for a few years, 100Mb/s connections are still pretty common. Moving a multiple gigabyte restore is kind of like pushing rush hour traffic leaving a big city out over a single-lane dirt road.

I am working with the campus print shop at the university where I work to help them select a software package that will allow them to accept online orders while streamlining their internal processes. We were just discussing storage and backup needs and one of our system administrators calculated that it would take nearly 24 hours to do a full backup of server space they are requesting we provide. Obviously we'll need to figure something else out for how we manage that.

Frank makes some good points about how to manage your customers. Spread them out over several servers and keep enough capacity open on your servers so that you can move a few customers on to each in the event that one server goes down. Though this may cost you a little more, it will gain you more in reliability and customer satisfaction when problems do occur. What's more expensive: two half-full VPSes or losing all of your customers on one VPS because it was down?

--Jason

[atechstl]

I know that I can create an Image of the Server in Virtuozzo but I am only limited to one. These servers being virtual is there not a SAN taking snapshots of them? If things are stored local than I would expect there should be local backups available to prevent this time delay.