We have multiple dedicated servers but after some years of use disks started failing. Our provider (100Tb) changes failing disks. In fact data center changes these (Softlayer) as 100Tb only forwards them the requests, because we have the servers in Softlayer facilities. Problem we encountered was that provider changed old failing disks with other old failing disks. This is the 3rd time we received old failing hardware as replacement but we checked and got it replaced. Support is responsive but there are probably some management flaws (probably they recycle disks without verification, which in time results in a pool of bad hardware causing bad services). Recovering data and restoring involves a lot of hassle. Currently 2 of our servers are offline and clients are complaining. One disk was just changed 1-2 months ago but the "new" one was also a bad one. Now we have to move data again to another drive. We check every time for SMART errors because we always seem to get failing drives. Now we got 2 drives that are old (13 000 / 7 000 power on hours) but at least these don't show smart errors. Do you have experience with other data centers? Does anyone guarantee replacement with new hardware that is... new? PS: If you want to check your disk stats on centos and don't know how: yum install smartmontools (installs tool if you don't have it) smartctl -A /dev/sda (shows under RAW VALUE stats like Power_On_Hours; and errors that show disk is failing Raw_Read_Error_Rate, Current_Pending_Sector, Reallocated_Event_Count )
I don't know of any company specifically that guarantees new hardware. There is no reason for a host to put brand new hard drives in a server, just as there is no reason for a host to put a new CPU, or RAM sticks, or RAID controller, or anything else. There's just no need for it. Most datacenters usually have a mess of parts laying around, and they use whatever they can to build the server. If a hard drive is still in working condition, there's no need to throw it out and buy a brand new one, if you have 100 of the exact hard drive laying around collecting dust. That said, they shouldn't be giving you damaged hard drives. If you search enough over on WHT (I see you posted there as well), you'll notice you aren't the only one who has had these issues with SoftLayer. Quite honestly, I think they are incompetent, and have been since ThePlanet went away. We used to have servers with them, but as soon as SL came in, things went downhill and we jumped ship almost immediately. Bad drives should be thrown in the garbage, and if that isn't happening, it is simply a matter of incompetence on somebody's behalf. Your best bet is going with the latest technology. If you can get away with SSD drives for instance (most people can't due to the lost space), chances are the drives will be new, as most hosts/datacenters don't have a bunch of old ones just sitting around. The technology is new enough that most hosts are going to be ordering them new.
I agree that there's no need to replace hardware with new drives each time but you have to have a cut-off point for re-cycling drives. Our cut-off point for customer servers in 4500 hours (just over 6 months) with no errors. We will recycle drives older than that for our own servers such as DNS, or FTP, but never for a customers dedicated server. 13,000 hours is a lot of hours (1.5 years) and I wouldn't be comfortable re-using them except as major emergency stand-by drives. Drives are pretty cheap when bought in bulk, and surely SL are buying in a bigger bulk than most companies so will be getting their drives for a massive discount? Each to their own.