Comments on: The Software RAID Inflection Point https://smbitjournal.com/2016/12/the-software-raid-inflection-point/ The Information Technology Resource for Small Business Sun, 12 Feb 2017 15:22:54 +0000 hourly 1 https://wordpress.org/?v=6.9.1 By: Scott Alan Miller https://smbitjournal.com/2016/12/the-software-raid-inflection-point/comment-page-1/#comment-25943 Wed, 01 Feb 2017 00:53:46 +0000 http://www.smbitjournal.com/?p=1009#comment-25943 In reply to Dashrender.

That depends, for normal operations it matters very little. Both are very fast and this is why we generally don’t worry about which you use, throughput is really determined by the drives and not the RAID controller. Faster, more powerful CPUs will drop the latency just a tiny bit. What has more affect is the ability to have large caches using system memory with software RAID. During resilver operations the power of the CPU might be felt to keep repair times down, especially if there is drive access happening during the resilver. Hardware controllers can be easily overrun by large SSD arrays whereas the system CPU can handle vastly more. So it depends very much on the specific circumstances.

]]>
By: Dashrender https://smbitjournal.com/2016/12/the-software-raid-inflection-point/comment-page-1/#comment-25942 Tue, 31 Jan 2017 23:53:19 +0000 http://www.smbitjournal.com/?p=1009#comment-25942 While those of us that have had discussions with the author know that hardware RAID gives most of it’s advantages today to blind swapping, my question for the author is:

Have processors on RAID controllers at least reached parity performance-wise for the storage subsystem versus using the system CPU?

]]>