Skip to contentSkip to conversation

How SRP initiator performance has been improved via scsi-mq

LSF/MM TOPIC Bart Van Assche 2 participants 2 messages Jan 14, 2015 – Jan 15, 2015 SCSI

lore
Share & tools
lore
#0 lore raw

Recently it has been proposed to discuss how to add scsi-mq support in the iSER initiator during the 2015 edition of the LSF/MM summit. In the discussion that followed that proposal several questions were asked about how scsi-mq support has been added in the SRP initiator driver, which design choices have been made and which lessons have been learned. Hence the proposal to add a topic to the LSF/MM schedule for reporting how scsi-mq support has been added in the SRP initiator.

For the actual discussion see also Sagi Grimberg, [LSF/MM TOPIC] iSCSI MQ adoption via MCS discussion, linux-scsi mailing list, January 7, 2015
(http://thread.gmane.org/gmane.linux.scsi/98199).

Christoph and Mike, if you would like to complement this topic with a presentation about the scsi-mq core itself and/or integration of multiqueue support in the device mapper multipath driver, I would welcome this.

Bart.

#1 lore raw
Quoted 8 lines · Bart Van Assche

Recently it has been proposed to discuss how to add scsi-mq support in the iSER initiator during the 2015 edition of the LSF/MM summit. In the discussion that followed that proposal several questions were asked about how scsi-mq support has been added in the SRP initiator driver, which design choices have been made and which lessons have been learned. Hence the proposal to add a topic to the LSF/MM schedule for reporting how scsi-mq support has been added in the SRP initiator.

Quoted 8 lines

For the actual discussion see also Sagi Grimberg, [LSF/MM TOPIC] iSCSI MQ adoption via MCS discussion, linux-scsi mailing list, January 7, 2015
(http://thread.gmane.org/gmane.linux.scsi/98199).

Christoph and Mike, if you would like to complement this topic with a presentation about the scsi-mq core itself and/or integration of multiqueue support in the device mapper multipath driver, I would welcome this.

A big long blk-mq / scsi-mq sessions sounds like a good plan. That can include a brief overview of the existing work and all the still pending bits for iSCSI/iSER, multipath and so on.