[GTALUG] Swap and Hibernate on SSDs
Russell
rreiter91 at gmail.com
Sat Jan 27 10:09:33 EST 2018
On January 27, 2018 8:49:15 AM EST, Giles Orr via talk <talk at gtalug.org> wrote:
>In the early days of SSDs, there was a lot of talk in the Linux
>community
>that you shouldn't have a swap partition on an SSD. An indirect side
>effect of that is that it becomes impossible to suspend-to-disk (aka
>"hibernate") on a SSD-only system. Although I suppose that you could
>have
>the swap partition and set swappiness to 0. So I have a new SSD
>system,
>and was wondering the current state of these things:
I'm still tweaking/breaking (in) my new SSD desktop system myself. I took Hugh's advice and did some research on TRIM.
This link has lots of info.
https://wiki.debian.org/SSDOptimization
My observations so far.
>
>- is it considered okay to have a swap partition on an SSD drive?
Yes, it's getting to be an inevitability. The key is to optimize the system to minimize the number of writes. ie. config w/ noatime to reduce writing all of the files access times to the drive. Some questions around that tho. Later kernels use relatime as a default middle ground.
Kernels 4+ have lazyatime available.
https://lwn.net/Articles/621046/
>- how should I set the swappiness?
That all depends on the operational stress factors. I think that on an average desktop, the default is fine.
https://lwn.net/Articles/83588/
>- is it considered okay to hibernate to an SSD?
Just from what I've read so far, if you keep to the convention of 2x(ram) for the swap you shouldn't have any problems.
>
Again, from what I've read recently, the general feeling is that an enterprise SSD, capable of >50gb writes in a day, is as virtually stable as a standard HDD. Consumer SSD's are considered the same if writes =<20gb a day.
HTH
>--
>Giles
>https://www.gilesorr.com/
>gilesorr at gmail.com
--
Russell
More information about the talk
mailing list