[gpfsug-discuss] Gpfs Standar and JBOD enclosures

Andrew Beattie abeattie at au1.ibm.com
Mon Mar 9 12:09:48 GMT 2020


Iban,

Spectrum scale native raid will not be an option for your scenario.

Scale erasure code edition does not support JBOD enclosures at this point.

And Scale Native RAID is only certified for specific hardware ( IBM ESS or Lenovo GSS )

This means you either need to use Hardware raid controllers or a shared nothing disk architecture ( I would recommend the raid controllers )

Regards, 

Andrew

Sent from my iPhone

> On 9 Mar 2020, at 21:59, Iban Cabrillo <cabrillo at ifca.unican.es> wrote:
> 
> 
> Dear,
>    We have a cluster with more than 1PB running Spectrum Scale Standar 4.2.3-9 version.
> 
>     The filesystem that is running now is configured using a DAS systems with NSDs over a NetApp DDP (distributed raid 6).
>     These days have arrived to our hands a pait of JBOB (D3284) enclosures and 4 new servers, the idea is connect it in pairs by two 2 x (1 enclosure + 2 SAS servers).
>     Is there any way to add this enclosures to the actual Cluster using GNR (something similar) o this option is only available for GSS/ESS?
>     If GNR is not an option, Is there any other way to add this enclosures to the cluster keeping the redundacy (path/server)? Maybe one failure group by enclosure with data replica set to 2 (we will lost the 50% of the sapace)?
> 
> Regards, I
> 
>    
> 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://gpfsug.org/pipermail/gpfsug-discuss_gpfsug.org/attachments/20200309/f9b4ebf7/attachment-0002.htm>


More information about the gpfsug-discuss mailing list