[rsyslog] [OT] Re: reliability of SSD disks?

Aaron Wiebe epiphani at gmail.com
Wed Aug 19 16:09:00 CEST 2009


On Wed, Aug 19, 2009 at 9:52 AM, Mr.
Demeanour<mrdemeanour at jackpot.uk.net> wrote:
>
> So the website (http://www.fusionio.com) doesn't seem to offer any
> information on the number of write cycles these devices are able to
> sustain. Since they are NAND Flash, this would be expected to be up to
> about a million, rendering them of dubious value as a part of a logging
> system. With wear-levelling, once cells start dying, I'd expect the
> death-rate to climb very rapidly indeed.

This hasn't been true for years.  With the improvements in NAND and
wear leveling techniques, the performance and lifetime of NAND based
storage outweighs that of spinning rust.

On fusion-io specifically, they also use a byte-level RAID type
algorithm to handle failures.  In their talk earlier this week, they
told us that they have -never- had a card come back failed.  But many
people still software-raid1 two cards together.

DRAM tends to be more complicated in power handling, and getting 640GB
of it can be very expensive indeed.  Granted, the applications are
similar, but I would expect PCIe based devices to take bigger hold
over the next few years.

-Aaron



More information about the rsyslog mailing list