Gluster users - Dec 2014 - two questions regarding glusterfs cluster: tierd storage and hotspare brick

hi, all,
    We are using gluster 3.5.2 on standard 3TB*12 machines, some hosts also
got SSD on the machines as well;
two questions:
1. Is there mature solution, I can use the SSD as the write cache for sata
disks? So I can combine the power of SSD as write cache(and even read
cache), and the capacity provided by SATA disks?
    I heard on Ceph there is  tied storage solution already, on glusterfs i
did't see that on documentation;
    Anyone used solution like facebook flashcache solution?

2. As sata disks fails often(relatively, when you got lots of disks...),
each time a disk fails, i will have to manually replace the disks, and
replace the bricks;
    Can we setup spare bricks?
     Say, I have 12*3TB sata disks, I only use 10 disks as 10 Bricks with
data on it, and have 2 disks configured with filesystem, as part of volume,
without putting data on the brick/filesystem? So in case on brick fails,
this brick automatically replace that failed disks? Just like hotspare in
raid?


Thanks;



-- 
Regards
Zhu Chao
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://www.gluster.org/pipermail/gluster-users/attachments/20141228/bc5ed8d2/attachment.html>

________________________________


2. As sata disks fails often(relatively, when you got lots of disks...), each
time a disk fails, i will have to manually replace the disks, and replace the
bricks;

Your disks should be part of a RAID set. (you choose the RAID level!)
If a disk fails, the RAID controller will keep the RAID running, and will
rebuild when a new disk is inserted.
Or you can have hot spare disks of course.

But yes - unless you have a robot arm in the server room you will have to
manually change the failed disks!




***********************************Viglen***********************************
Viglen Ltd, Registered in England No 1208441. Registered Office: 7, Handley Page
Way, Colney Street, St. Albans, Hertfordshire AL2 2DQ.

Information in this electronic mail message is confidential and may be legally
privileged. It is intended solely for the addressee. Access to this message by
anyone else is unauthorised. If you are not the intended recipient any use,
disclosure, copying or distribution of this message is prohibited and may be
unlawful. When addressed to our customers, any information contained in this
message is subject to Viglen Terms & Conditions. Please rely on your own
virus checker and procedures with regard to any attachment to this message.
***********************************Viglen***********************************

______________________________________________________________________
This email has been scanned by the Symantec Email Security.cloud service.
For more information please visit http://www.symanteccloud.com
______________________________________________________________________
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://www.gluster.org/pipermail/gluster-users/attachments/20141229/f86b778a/attachment.html>

On 12/28/2014 02:07 PM, Zhu,Chao wrote:
> hi, all,
>      We are using gluster 3.5.2 on standard 3TB*12 machines, some hosts
> also got SSD on the machines as well;
> two questions:
> 1. Is there mature solution, I can use the SSD as the write cache for
> sata disks? So I can combine the power of SSD as write cache(and even
> read cache), and the capacity provided by SATA disks?
>
Data tiering feature being developed for 3.7 can help here. More details 
about tiering can be found in this larger proposal for data 
classification [1].
> 2. As sata disks fails often(relatively, when you got lots of disks...),
> each time a disk fails, i will have to manually replace the disks, and
> replace the bricks;
>      Can we setup spare bricks?
>       Say, I have 12*3TB sata disks, I only use 10 disks as 10 Bricks
> with data on it, and have 2 disks configured with filesystem, as part of
> volume, without putting data on the brick/filesystem? So in case on
> brick fails, this brick automatically replace that failed disks? Just
> like hotspare in raid?
There's no automatic mechanism to do this from the CLI today. GlusterFS 
has the ability to detect brick failures [2] and also to perform an 
online replacement of a brick in a volume. By combining these two 
features and defining a mechanism in management to add/remove spare 
bricks, this functionality can be achieved.

Regards,
Vijay

[1] 
http://www.gluster.org/community/documentation/index.php/Features/data-classification

[2] 
http://www.gluster.org/community/documentation/index.php/Features/Brick_Failure_Detection