|
That's a good start, but it doesn't come close to identifying the main benefits and advantages and why you should care as a customer... after all, SAS disks are still considerably more expensive than SATAII, so there's a very good chance that hosting providers using these in their servers will want to charge you more, or are otherwise forced to cut corners somewhere compared to their competitors.
Previously there were IDE (PATA) disks which were often used in low cost computing applications such as desktops etc., and SCSI disks offering higher data transfer speeds (at a price!) often used in servers.
Although this is primarily about the interface rather than the disk itself, it's worth noting that the manufacturing and testing processes used are appropriate to the implied applications of each disk type - therefore SCSI disks offer a longer MTTF compared to their PATA friends.
Fast forward to the world as we know it today, and SATA is pretty much the successor of PATA, and SAS the successor of SCSI. The intended use of each "family" is therefore broadly the same: SATA for low(er) cost, and SAS for enterprise.
Although current technology allows SAS and SATA(II) to achieve similiar data transfer speeds, the SAS interface has a predefined roadmap leading it up to 1,200MB/s (IIRC) compared to the current 300MB/s achieved from both. I think SAS also generally has quicker seek times, although not certain if this is necessarily a function of the "interface/disk" as such or just simply because SAS disks usually have a faster spin speed (rpm).
Since SAS disks are designed and manufacturered to be used in servers, they are actually intended for 24/7/365 operation and true enterprise use. This means that you get longer MTTF (as above) but also an improved BER (bit error rate). The BER is how frequently you will encounter a non-recoverable bit when trying to read from the disk: it's specified by the manufacturer along with the various other technical specifications for each disk.
However, some typical numbers I found regarding this indicate that SATA disks may have something like 1 bit error per 11.6TB read. In contrast, typical for SAS disks would be 1 bit error per 1,164TB read.
What does this mean? Well, it's common to run some configuration of RAID array to help protect against the problems disk failures cause (and in particular, the lower MTTF of SATA, but it's still just good sense for SAS users too!). To rebuild the RAID array in the event of a failure (and assuming that the RAID configuration is one supporting rebuild such as mirrored/parity bit etc. - i.e. not just striped, as then you have serious data loss whatever!) generally requires that you have an error free copy of the data available on the other disk(s). If you get just 1 error in a single sector, that corresponding sector on the replacement drive cannot be reconstructed!
So, in this case it's great that Simplehelix have seen the light and upgraded - but a tad worrying that they haven't used it all along for the above reasons.
__________________
Please login or register to view this content. Registration is FREE :: DDS & Dedicated, UK & USA-based Please login or register to view this content. Registration is FREE, Reseller & Shared Hosting
Experienced Parallels Platinum Partners (Plesk since 2001, Virtuozzo since 2003)
|
