Hey everyone, we are getting more and more requests these days for Disaster Recovery (DR) options and backup solutions whether it be basic geographically separated DNS servers or full blown iSCSI/SAN’s across two independent networks. I’m wondering if others have seen this increasing trend and considered the importance of DR solutions. It would seem as more businesses realise how dependent they are on technology, protecting their services and data is becoming a very important issue.

Regards James

Hi James,

Yes have to agree with you. We are getting more and more customers requesting diverse locations. I feel it is a sign of the maturing of the market, and a good thing too.

Sites needing high uptime's realise that no matter what a DC provides re SLA's if they really need 100% availability then having more than one server in more than one location just makes sense.
I'm also seeing some customers looking to co-lo equipment as a DR solution for their own inhouse computer room. Makes sense, as that way if their building becomes un-tennantable they can send some key staff home to work remotely, connecting into the DR site.

Cheers,

Matt

Yes the market maturing is a very good move in the right direction. As a example we had a client who required us to monitor 24/7 temperatures of fridges for food safety regulations, not only did he co-locate his servers which collects the data from his onsite temperature monitoring devices, he also installed redundant internet feeds (ADSL and Cable) to insure his servers can always connect to his equipment. At first one would think “over kill” but when it comes to food safety regulations you just can’t afford to not follow the rules.

Regards James

Absolutely - at least for corporate hosting cilents. The expectations are growing higher with every passing day. Also you will not pass a Big 4 (Deloitte, PWC, EY, KPMG) risk management audit which quite a few clients demand these days.

We are currently addressing this by building a new POP in a different state and doing SAN to SAN replication between the two sites. It is a big investment and increases your operational expenses, but a great technical solution.

I think you really do need a 'warm' DR site for your DR plan to be useful. Having offsite tapes and a few servers in your office will not realistically help you recover your service (it could still take you 5+ days to buy everything you need and get it all back up and running). You really need a full set of infrastructure as per your primary site (i.e. redundant routers, switches, firewalls, domain controllers). If you're going to buy all this hardware you might as well keep it running in a DC in another state and have it ready to take over if your primary site fails.

Umm.. When has DR NOT been important?

I think everyone knows it's important but love to just brush it under the carpet as an 'unseen' possibility. These sort of organisations generally change their tune when their entire primary location goes down (or they lose all their data).

We've got a LOT of customers both using us for DR and for their live stuff (in geographically separate & diverse fibered datacentres). I honestly can't tell you I've ever been in a situation where DR wasn't talked about and considered.

Maturing of the market? Nah, I don't think that's it, I think it's more about the media shakeup caused by things like British government departments losing DVD's of personal data and companies in the U.S. having to disclose vulnerabilities. Nowadays if a large company was to say they lost their data and they didn't have a DR site (which as someone already reported would mean they fail an audit) the media would have a field day and people would be bailing out quick smart.

Stu

This has been the case for the last 8 years at least.. I think its an increase for your sort of client base

No really it depends what market you target.. Plenty of our clients think about DR more than the main operation or at least equal amount of time and sometimes more money is spent on DR.

We run most operations from Equinix and doing some DR at GS.

Bandwidth between is so cheap at the moment the way we do it.

Yeah, everyone's looking for data centre space nowadays.

Oh yeah, we're currently doing an internal refresh of our DR and associated storage strategy. Once you throw in block level replication licensing your DR component is actually ~30-40% more expensive than the original implementation.

Stu

Are you able to disclose how you are able to do it so cheaply? We are moving into GS and may want to setup a smaller POP as well at EQX.

Can't you just buy a 100Mb VLAN off Uecomm or PIPE Networks? Its around $4k + cross connection fees last time I looked...

It's a lot cheaper than that last time I checked. More like half that.

No doubt, that was last time I looked. Which was 3 years ago.

DR has always been important to the bigger hosts. Bear in mind though that most of the smaller hosts don't do proper backups let alone any DR. I'd say the percentage that would be covered if they lost an entire datacentre would be about 50%.

Also, it depends on what you mean by DR, there are about a hundred different types of solution ranging from RAID-1 through live replicated systems in 10 different continents. Not that I'd consider RAID-1 to be DR, but there are some that apparently do!!