Re: tgt and krbd

[Steffen W Sørensen <stefws@xxxxxx>]

On 06/03/2015, at 22.47, Jake Young <jak3kaj@xxxxxxxxx> wrote:

> I wish there was a way to incorporate a local cache device into tgt with librbd backends.

What about a ram disk device like rapid disk+cache in front of your rbd block device

http://www.rapiddisk.org/?page_id=15#rapiddisk

/Steffen

I could try that in my VM to prototype the solution before I buy hardware. 

Note this from given URL:

'Do not use this in a virtual guest or with a loopback device. You will not see any performance improvements for reasons I do not feel like explaining at the moment. In fact, the performance will be worse in such environments. Only use this with an actual physical disk device.'

RAM based cache is pretty dangerous for this application. If I reboot the VM and don't disconnect the initiators, there would most likely be data corruption, or at the very least data loss. 

I know, it's a trade of between perf vs integrity, but also note this:

'In RapidCache, all writes are cached to a rxdsk volume but also written to disk immediately. All disk reads are cached.'

'Enable general block device caching for: (1) Locally attached disk devices and (2) Remotely attached disk devices mapped over a Storage Area Network (SAN).'

Haven't utilized such my self yet. I'm still wondering about the difference between rapid cache vs normal linux page cache.

Believe rapid cache might ack write faster to an app and then handle the write to spindle following, and again looking at authors numbers write seems identical to spindle numbers so maybe write ack isn't given before staged to underlying spindle. Need to dig deeper into rapid cache.

/Steffen

Attachment: signature.asc
Description: Message signed with OpenPGP using GPGMail

_______________________________________________
ceph-users mailing list
ceph-users@xxxxxxxxxxxxxx
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com