Re: JBOD / ZFS / Flash backed

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On 09/04/18 19:00, Vincent Royer wrote:
Yes the flash-backed RAID cards use a super-capacitor to backup the flash cache.  You have a choice of flash module sizes to include on the card.   The card supports RAID modes as well as JBOD. 

I do not know if Gluster can make use of battery-backed flash-based Cache when the disks are presented by the RAID card in JBOD.   The Hardware vendor asked "Do you know if Gluster makes use of flash-cache in JBOD?"

If it does, I'm not certain how the size of this flash cache affects the operation. 


I thought you wanted to use ZFS underneath Gluster? You need to state your use case properly so we can help you.

Pretty much any RAID card I've come across does not use the on-board cache (which is always RAM, backup up with a BBU or a Flash chip(s)+Supercaps) when set to JBOD mode. You could use a RAID controller with GlusterFS with XFS or EXT4 underneath but it why not just use the redundancy built into GlusterFS? You could still use software RAID if you're really concerned about your data.

If you're using ZFS you don't even want to use a RAID firmware/card set to JBOD mode, you need an HBA or a dual-purpose RAID/HBA card running the HBA firmware (IT-mode in LSI/Avago/Broadcom cards, eg IBM M1015, 1115).

Even if you use XFS or EXT4 underneath Gluster I'd still look at leaving out a RAID capable controller, as if you can't get the same model and the card fails you can't just plonk the drives into any other box with SATA/SAS ports and just carry on as before.

In either case, don't use desktop drives as they often lie about if they flush their own RAM cache. Use nearline enterprise SATA or SAS drives.

Part of the point of GlusterFS and ZFS is it's "software defined", you use fast but dumb drive controllers so you don't have to ever worry again about hardware compatibility and availability, it's all in the OS/FS stack and the hardware conforms to open standards.

Eg:

Client apps > GlusterFS > ZFS | > HBA > JBOD
Client apps > GlusterFS> XFS | > HBA > JBOD

All the caching, resilience, failover is handled above the place where I've put the pipe character. This means your HBA and enclosures can go up in smoke, as long as you still have the drives you'll have your data.

What are your plans WRT your underlying brick FS? ZFS or other?






--
This message is intended only for the addressee and may contain
confidential information. Unless you are that person, you may not
disclose its contents or use it in any way and are requested to delete
the message along with any attachments and notify us immediately.
This email is not intended to, nor should it be taken to, constitute advice.
The information provided is correct to our knowledge & belief and must not
be used as a substitute for obtaining tax, regulatory, investment, legal or
any other appropriate advice.

"Transact" is operated by Integrated Financial Arrangements Ltd.
29 Clement's Lane, London EC4N 7AE. Tel: (020) 7608 4900 Fax: (020) 7608 5300.
(Registered office: as above; Registered in England and Wales under
number: 3727592). Authorised and regulated by the Financial Conduct
Authority (entered on the Financial Services Register; no. 190856).

_______________________________________________
Gluster-users mailing list
Gluster-users@xxxxxxxxxxx
http://lists.gluster.org/mailman/listinfo/gluster-users

[Index of Archives]     [Gluster Development]     [Linux Filesytems Development]     [Linux ARM Kernel]     [Linux ARM]     [Linux Omap]     [Fedora ARM]     [IETF Annouce]     [Bugtraq]     [Linux OMAP]     [Linux MIPS]     [eCos]     [Asterisk Internet PBX]     [Linux API]

  Powered by Linux