You are here

Home » Blogs » Konstantin Boyandin's blog

VPS as a high-risk hosting


What's a good hoster?

True friends are those who stay with you in a time of need.

True hosters are those who do everything possible to keep you as a customer. First of all, of course, by keeping high uptime and preventing disasters.

It grieves me to say a server hosting some of JaguarPC's VPSes went down on 30-th of September, 2012. Today, almost 17 days after that moment, the situation isn't fully handled. You can read further on this WHT thread, and on JaguarPC forum as well. I sympathize with everyone, hoster included. Hardware will eventually break down. The scale of disaster can't be predicted in most cases, even though the first traits of the upcoming failure should have been seen long before.

So, a good hoster is the one that handles such things as quickly and thoroughly as possible. I suppose JaguarPC will finally do that, since currently it's a matter of creating VPSes anew and copying whatever was backed up.

However, I am somewhat surprised to see many a customer just sitting and grieving about data lost, doing nothing but waiting. In my opinion, every person choosing a VPS (cloud server etc. — they do not make much difference in terms of recovering after major faults) should understand that even if VPS is managed, it will require exactly the same amount of time to recover if everything fails.

And, what's important, «managed» doesn't mean «buy and forget». The hoster will handle most of maintenance, but those famous chunks of data named backups remain the responsibility of the customer.

If they think otherwise, it's only a matter of time when a major hardware (or other) failure proves they were wrong.

Recovery plan

What will I do if it becomes clear that the current server, where this and some other sites are hosted, failed beyond recognition and will unlikely return back unscathed?

If I feel the recovery is in progress with no ETA, I will switch to another VPS as soon as possible. Yes, it means keeping a copy of an VPS, fully synced, with exactly the same software installed. All I need to do is to run some scripts, to restore some backups (it's a quick task) and change DNS records. In the worst scenario, within 6-18 hours I will have all the sites alive again, with a day-old backups restored.

If you think that's too expensive, to keep a copy of a VPS, answer two questions:

  • exactly how much downtime is required so that your business started to suffer grave losses?
  • how much resources, money included, are you ready to invest into a backup VPS?

I mentioned several times CloudSigma, Amazon AWS, other providers that allow keeping «dormant» servers, periodically awakening them to sync with main VPS and putting to sleep again. They do not cost really much and they could be a good means to save you from long downtime periods. Actually, if you are not very short of funds, you can just keep the backup VPS running permanently, using load balancing to choose between two replica.

Much time to set up? Generally speaking, yes. Especially for the first time.

Much expenses to maintain? Not that especially much, but yes, you will have to find someone to keep an eye on the setup, to bring your other server online in case of disaster.

Just decide, what is easier to afford — a probable complete loss of your site(s), perhaps with several last weeks of activity unrecoverable, or certain additional expenses to make sure you can survive a crash of any VPS without losing virtually everything.

In case you are eager to know what I usually do to secure a newly-created VPS, stay tuned.