Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
On Sat, Feb 11, 2012 at 7:30 PM, Pascal BERTON <pascal.berton3 at free.fr> wrote: > Florian, > > I've watched your video, very interesting indeed... Unfortunately for you it > raised some more questions to me :-) > First, just to be sure I correctly understood : flashcache will only (But > significantly) improve read operations, no impact or so on writes right ? It's a write-back capable device, so if you configure it in that mode, of course it will have an impact on writes. > Then, a year or 2 ago I had the opportunity to play around with Ming Zhao's > dm-cache (completely apart from DRBD by the way, it was a standalone > system), and to what I remember it roughly looked like this flashcache thing > I didn't know of, except it makes use of a ramdisk instead of an SSD drive > (And it has an impact on write operations, as far as I remember). I > immediately see one advantage compared to flash-cache : it would (should?) > be more performant, being memory-based instead of IO-based. However, > inconvenient, it would be certainly smaller. > Have you ever had the chance to add that dm-cache feature to DRBD the way > you did it with flashcache ? Do you know/think it would work efficiently > with DRBD ? And, lastly, do you think it would make sense to think of > combining dm-cache and flashcache together to get kind of an L1+L2 storage > cache level pair ? As far as I'm informed there are currently three generic block-layer caching devices available on Linux: flashcache, dm-cache and bcache. Of those, only bcache seemed to be seriously interested in a push towards mainline until recently (though that may be changing). As for myself I have zero experience with bcache and dm-cache, so flashcache is the only one I can comment on. Hope this helps. What's your intended use case? Cheers, Florian -- Need help with High Availability? http://www.hastexo.com/now