Well it seems that for the last 3 days, a couple of comapnies in this forum have said that there Hard Drives have failed at the Data Center.
The first comapny took over 35+ to recover, and clients had to do reset acounts manually.
The second is at 28 hours and going still.

So the question has come up as follows.

1) How long does it take to change out a Hard Drive and reload?

2) What is the exceptable time you would allow a faliure like this to be resoanble, before you split?

Experts and owners comments a plus on answering this. I like everyone to comment esp. if it has happened to you.

24 hours is not unreasonable. By the time you pull the old drive, test it, stick new one in, reformat and reload it, test it. If you're like us and you use Raids on all of your servers you also have to wait for the drives to fully resync.

Edited: It's why we use Raid drives and do daily and weekly tape back-up on all of our servers. In a worst case scenario we can reload everything on to a spare server.

IMHO a top notch DAC would have spare drives preloaded with the OS ready to go. Thus I would expect that a hardrive failure be recoverable in less than 24 hours. However this would depend upon the amount of data needing to be restored.

I thought that is what raid drives are for, one fails the others can keep going with no interuptions. Does this mean that these comapnies did not have raid drives installed?

sort of looks that way don't it.

RAID are costly and now when people are asking for lower prices, servers are no longer industrial strength real servers but of 'whitebox' variety. It works most of the time. I guess if you need better, you would have to pay more.

Originally posted by boywonder
I thought that is what raid drives are for, one fails the others can keep going with no interuptions. Does this mean that these comapnies did not have raid drives installed?


many don't simply because the prices they charge for hosting don't cover the cost to provide raid or tape back up. most will stick to a 2nd hard drive.

In any case, no matter what the host says they do for backup, you and your clients should always have your own backups. Never rely on your host.

Hell the money I would have paid for these last 3 days for a raid drive on the host server.

The new hosting provider I am looking for, I am asking that question? What is the plan in a failure.

Hell the money I would have paid for these last 3 days for a raid drive on the host server.

Well, most don't feel the same way before the problem occurs. You would probably see at least a 3 or 4 times hike in prices of hardware if a RAID-5 subsystem is installed. This would mean paying 4 to 5 times the current cost to support it.

Yes, when looking for a new host, check to see if they have a RAID subsystem. Be prepared to pay a lot more.

True, the onus for backups should ultimately still lies with the client. It is your data and it is your job to perform the necessary backups.

Or the host could have a backup server and just internally rsync their drives to this server.

Much cheaper than raid and it takes a few hours to get everything back (if no drives are available) or an hour if a drive is available.

Just my 2c.

Anything over 24hours is too much.

A quality backup (few terabytes of storage) server using rsync is not cheaper than RAID.

-Brendan

Raid costs more at startup but is cheaper in the long run. (like everything in this business)

Now, if people would just stop using these lame desktop (wdc) drives in their 'servers', I'm sure it would help with all these hard drive failures. :)

Now, if people would just stop using these lame desktop (wdc) drives in their 'servers', I'm sure it would help with all these hard drive failures

:agree: :beer:

...all to save a few bucks.....

So what are the good Data Centers now, that most people use dedicated servers with to rent out for reseller accounts?

One thing you could do is opt for a colocation plan and build the RAID-5 into the system. This way you do not need to pay for additional monthly charges for the RAID implementation and you could configure the server as you like.

Changing a hard drive takes 15 - 25 (MAX) if you are on a high end provider then it should be done within 1 hour but low end is fine within 6 hrs, thats just my oppinion ATM

If I know my host hasn't got raid 5 and a daily backups implemented, I'll start looking for a different host. You don't use only hardware raid, there also exists implementations of software-oriented raids. These make the system a little bit slow but are very good for data security.

Or the host could have a backup server and just internally rsync their drives to this server.

If you're actually making use of this strategy, I'm surprised that none of your users have called you on it. Without running rsync at intervals that would seriously degrade server performance, I can imagine this wreaking lots of havoc---especially for users who have a lot of dynamic content.

I'd be willing to consider this as a last-resort in case of a server failure, but I'd never consider it a normal recovery from a failure.

I have started to read into Dedicated and Co-location. I am not sure what they are in detail? Also I know I could reselle the extra soace I do not use. I am not sure what I would need to run a co-location or dedicated.

One thing for sure, great support and back-up plan in place if something failed.

Well... Raid doesn't necessarily means it's fast to recover. The rebuild process is anything but fast. Recently someone I used to host my junk (they are running Raid 5) lost one of the harddisk. They ran the rebuild process for a day or so... and in the end, we still didn't get our data back... (it's not 100% fail-safe)

sprintserve, I think your provider is pulling your leg...

The idea of RAID (with the exception of RAID 0) is that one (or in some cases, even more) drives can fail without any data loss, and with perhaps only a slight impact on system performance.

Done right, the only thing that can take out a RAID array is either a controller failure, or multiple drives failing at once. (Most controllers allow for 'hotswap' drives that can be brough tinto service after a failure, minimizing the risk of multiple drives failing in a short period of time.)