When did RAID became independent?

Once upon a time I was told about that cool technology that lets you take several hard drives and glue them “together” to a single big volume. This technology was called RAID – Redundant Array of Inexpensive Disks – and that it was. It brought us greater levels of reliability and performance – and it was inexpensive compared with other technologies and since hard drive prices are falling for years and storage space is growing along with that it’s getting even cheaper than anything else you could use to store data securely. Some of us even backup to a independent RAID system.

In the beginning of this all there were several hard drive interface technologies used – mainly it was Parallel ATA and SCSI. It was widely accepted that the SCSI drives are specified for 24/7 server usage and were almost everytime faster than their consumer PATA relatives. It was accepted that if you want to build a reliable industry grade RAID you would want to use SCSI drives – the SCSI bus system even had advantages like up to 7 drives per bus compared to just 2 drives with PATA or hot-swap capabilities.

Over the last years it turned out that SATA is the new interface technology that replaces the old SCSI and PATA. There are several server grade SATA drives available now – these drives are getting cheaper, faster and bigger by the minute. So there’s not a real purpose for anything “more server than server-SATA” you might think. Again if you want to build inexpensive and redundant storage arrays there is nothing cheaper than standard or even server SATA drives. They are fast, reliable and huge.

So some years ago the industry presents: the SAS interface. It’s called “Serial Attached SCSI” and is the “new cool thing in hard disk storage”. There are some niche features that may or may not justify the existence of SAS. A fact is that SAS hard drives of the same size and speed are more expensive.

“SATA is marketed as a general-purpose successor to Parallel ATA and is now common in the consumer market, while the more expensive SAS is marketed for critical server applications.(Wikipedia)

It’s getting worse: The industry started to offer fast hard drives (15000 rpm) only for the more expensive SAS interface. The few 15k rpm SATA drives are not slower in any way than their SAS versions – but they are not widely available and all of a sudden the same price like the SAS version.

But back to the definition of RAID:

So over the years the technology made a giant leap forward and all of a sudden you find yourself using very expensive hard drives while glueing them together to giant volumes (it’s now terabytes…petabytes…). While consumer hard drives are available for about a third (at least) the price of the server version of the same drive. It seems that the widely accepted definition of inexpensive is replaced by independence. I do know that there are use cases when you want to use the fastest spinning drive available regardless of the price – but I also think that there could be affordable fast spinning drives if we shouldn’t be bothered to pay the marketing-fee that SAS brings. It’s plain marketing to make new 15k rpm drives only available for SAS and not for SATA. Marketing and nothing more.

As it turns out many industry (marketing) brains (hey, even wikipedia) are switching to a new definition of RAID. It’s now a Redundant Array of Independent Disks – which I think is a definition that could not be worse. It’s not independence we gain with the new definition.

Source 1: http://en.wikipedia.org/wiki/RAID#cite_note-1
Source 2: http://en.wikipedia.org/wiki/Serial_Attached_SCSI

Comments are closed.