How does Enterprise storage work? Does the storage server use "regular" retail Hard Drives?

There are disks that are more special and are supposed to be better in some ways though I personally never really got an answer what is the real difference between an enterprise and consumer hdd. There are the obvious firmware differences and some features that are custom made for each specific storage vendor but these are usually not critical changes that affect performance. At the end the basics of the hdd are the same, the writing method to the disk is the same and so does the reading. There are differences in how the disk handles failures, it may retry them for only a short time (enterprise) or a much longer (consumer). I've heard claims of better bearings and better components for 24x7 operation rather than a consumer device expected to work a few hours every day but not idea if that is real and how much of a difference it really makes. One difference in the drives is the speed, some storage systems espouse the use of very high speed drives (15K RPM) that mean the capacity is reduced and others can take "normal" disks of 7200 RPM and provide very good performance with them. There are also different disk interfaces (FC, SAS, SATA) and while there is a big gap in reporting and monitoring abilities between FC/SAS and SATA the only difference I'm aware of in the data-path is error handling in case of soft-errors (such as media errors, SATA aborts all outstanding IOs where SAS at least only fail the offending IO). The storage system itself is the key point as and have said, it is designed and implemented to make the best use of the disks and their performance characteristic in ways that a normal system cannot. It touches topics of distributing data on different disks to optimize the load, arranging the data on the disk for best possible performance gains, caching a lot of data on RAM and possibly SSDs to reduce the need to reach the disks for that it also employs smart caching algorithms that take a different tradeoff compared to what the OS kernel will likely do. There are also differences in the level of testing that a disk undergoes before being shipped to a customer for enterprise storage. Even after the disk was tested at the vendor it goes further testing at the storage system vendor for a non-trivial period of time and under stressful loads to weed out the bad disks, and there are quite a few returned disks from that process. How the storage system works internally is not widely disclosed and it differs between systems even inside the same company (EMC VMAX vs. VNX, IBM DS8k vs lower DSes vs XIV).

It is usually not public which hard disks exactly are used in Enterprise Storage Arrays as from EMC or NetApp, but they are essentially re-branded normal hard disks. The value of an Enterprise Storage Array does not come from special hard disks, but features as a fast RAID, very intensive testing, and higher level features as snapshots, quotas, replication, providing file and block storage from the same hardware management UI, and much more.

Most disks be it FC, SSD, SATA or SAS pretty much look the same as the retailed Hard Disk Drives. But on pondering a little one can see small paddle cards attached to the drives for protocol compatibilities. Also drives have a plastic casing specifying drive type and capacity and this casing mostly covers up the actual drive vendors label and creates an easy pull-out push-in method to handle drives in an array. It is generally said that the mean time between failures MTBF's for these drives are very low due to superior quality of the components used and that along with the enhanced firmware makes these drives expensive. Vendors per se have an enterprise line up of drives that are far more superior in quality and better service than retail drives. Other than physical differences there are quite a few changes with the firmware. One can set up specific flags on the drive that would ensure better crc checks and eliminate bit errors amongst other features that can be enabled/disabled. Also these drives frequently get firmware upgrades eliminating any problems with drives themselves. These days the enterprise drives have been tuned to be so smart that one can predict the drives imminent failure based on previous events and get it replaced before any issue has been observed. I would go so far as to say that the enterprise drives are expensive only because of the support and extra features that will be at your disposal though the functionality is mostly the same, when you buy a premium product you buy the experience along with it and that costs.