Chapter 48. Storage

Multi-queue I/O scheduling for SCSI

Red Hat Enterprise Linux 7 includes a new multiple-queue I/O scheduling mechanism for block devices known as blk-mq. The scsi-mq package allows the Small Computer System Interface (SCSI) subsystem to make use of this new queuing mechanism. This functionality is provided as a Technology Preview and is not enabled by default. To enable it, add scsi_mod.use_blk_mq=Y to the kernel command line.
Although blk-mq is intended to offer improved performance, particularly for low-latency devices, it is not guaranteed to always provide better performance. In particular, in some cases, enabling scsi-mq can result in significantly worse performance, especially on systems with many CPUs. (BZ#1109348)

Targetd plug-in from the libStorageMgmt API

Since Red Hat Enterprise Linux 7.1, storage array management with libStorageMgmt, a storage array independent API, has been fully supported. The provided API is stable, consistent, and allows developers to programmatically manage different storage arrays and utilize the hardware-accelerated features provided. System administrators can also use libStorageMgmt to manually configure storage and to automate storage management tasks with the included command-line interface.
The Targetd plug-in is not fully supported and remains a Technology Preview. (BZ#1119909)

SCSI-MQ as a Technology Preview in the qla2xxx and lpfc drivers

The qla2xxx driver updated in Red Hat Enterprise Linux 7.4 can enable the use of SCSI-MQ (multiqueue) with the ql2xmqsupport=1 module parameter. The default value is 0 (disabled).
The SCSI-MQ functionality is provided as a Technology Preview when used with the qla2xxx or the lpfc drivers.
Note that a recent performance testing at Red Hat with async IO over Fibre Channel adapters using SCSI-MQ has shown significant performance degradation under certain conditions. (BZ#1414957)

NVMe/FC available as a Technology Preview in Qlogic adapters using the qla2xxx driver

The NVMe over Fibre Channel (NVMe/FC) transport type is available as a Technology Preview in Qlogic adapters using the qla2xxx driver.
NVMe/FC is an additional fabric transport type for the Nonvolatile Memory Express (NVMe) protocol, in addition to the Remote Direct Memory Access (RDMA) protocol that was previously introduced in Red Hat Enterprise Linux.
NVMe/FC provides a higher-performance, lower-latency I/O protocol over existing Fibre Channel infrastructure. This is especially important with solid-state storage arrays, because it allows the performance benefits of NVMe storage to be passed through the fabric transport, rather than being encapsulated in a different protocol, SCSI.
Note that since Red Hat Enterprise Linux 7.6, NVMe/FC is fully supported with Broadcom Emulex Fibre Channel 32Gbit adapters using the lpfc driver. See the restrictions listed in the New Features part. (BZ#1387768, BZ#1454386)