Sun Microsystems, Inc.  Sun System Handbook - ISO 4.1 October 2012 Internal/Partner Edition
   Home | Current Systems | Former STK Products | EOL Systems | Components | General Info | Search | Feedback

Asset ID: 1-72-1374509.1
Update Date:2012-05-30
Keywords:

Solution Type  Problem Resolution Sure

Solution  1374509.1 :   Sun Storage 7000 Unified Storage System: Known iSCSI performance limitations  


Related Items
  • Sun Storage 7310 Unified Storage System
  •  
  • Sun Storage 7410 Unified Storage System
  •  
  • Sun ZFS Storage 7120
  •  
  • Sun ZFS Storage 7420
  •  
  • Sun ZFS Storage 7320
  •  
  • Sun Storage 7110 Unified Storage System
  •  
  • Sun Storage 7210 Unified Storage System
  •  
Related Categories
  • PLA-Support>Sun Systems>DISK>NAS>SN-DK: 7xxx NAS
  •  
  • .Old GCS Categories>Sun Microsystems>Storage - Disk>Unified Storage
  •  




In this Document
Symptoms
Cause
Solution
 Pool layout
 ZIL (ZFS Intent Log)
 Volume block size
 Partition alignment
 TCP connections
 Number of iSCSI sessions per iqn target
References


Applies to:

Sun Storage 7210 Unified Storage System - Version Not Applicable to Not Applicable [Release N/A]
Sun Storage 7310 Unified Storage System - Version Not Applicable to Not Applicable [Release N/A]
Sun Storage 7410 Unified Storage System - Version Not Applicable to Not Applicable [Release N/A]
Sun ZFS Storage 7420 - Version Not Applicable to Not Applicable [Release N/A]
Sun ZFS Storage 7120 - Version Not Applicable to Not Applicable [Release N/A]
7000 Appliance OS (Fishworks)
NAS head revision : [not dependent]
BIOS revision : [not dependent]
ILOM revision : [not dependent]
JBODs Model : [not dependent]
CLUSTER related : [not dependent]


Symptoms

In some circumstances, some iSCSI latencies might be discernible on iSCSI client side. This document provides some typical checks for iSCSI performance issues.

To discuss this information further with Oracle experts and industry peers, we encourage you to review, join or start a discussion in the My Oracle Support Community - 7000 Series ZFS Appliances

Cause

Among many possible causes, the following are of interest :

  • pool layout of datapool selected(raidz2, mirror, stripe)
  • availability of intent log devices (ZIL)
  • volblock size of the volume
  • partition alignment from client side
  • TCP connections
  • Number of iSCSI sessions per iqn target

Solution

Pool layout

To be able to provide a good level of IOPs, the pool layout to use is mirror. Use raidz2 only if the IO workload is similar to video stream or related to big file access.

ZIL (ZFS Intent Log)

    • All zpools have a ZIL even if a specific log device is not defined. If there is no dedicated log device, the ZIL is made from the data area of a pool. This can double the number of actual writes. A sync write would first be written to the ZIL on the drives and then commited to the normal data area. This "two for one" writes can have a extremely large impact on performance.To avoid this "two for one" penalty it is a good idea to separate the ZIL from the data drives and put it on a device with higher performance capabilities. A fast SSD is used for this purpose on the 7000 arrays : the logzilla.
    • If the host uses the iSCSI LUNs in ZFS volumes, the write cache can be enabled on the NAS head side. Thanks to this, the SCSI ACK will be sent back immediately to the host at high response time. If the host uses write cache but not through ZFS volumes, there is a risk of data loss in case of power loss on the NAS head. If integrity is a priority, always use logzilla. For better results, use 2 logzilla per pool (at least) in a stripe layout.

Volume block size

The volblocksize specifies the block size of a volume (iSCSI). The block size cannot be changed once a volume has been written, so set the block size at volume creation time. The default block size for volumes is 8 Kbytes. It can be set up to 128Kbytes.
Check what is the IO block size on host side before creating any iSCSI LUN.
Even if the VmWare block size is 1MB minimum, the genuine IO blocksize from the embedded application may be smaller.

Partition alignment

This topic is detailed on a blog from David Lutz.
With proper alignment, a single client block that is the same size or less as the volume block size of a LUN will be contained entirely within a single volume block in the LUN. Without proper alignment, that same client block may span multiple volume blocks in the LUN. That could result in 2 appliance reads for a single client read, and 2 appliance reads plus 2 appliance writes for a single client write. This will obviously have a big impact on performance if ignored.
For details, see Partition Alignment Guidelines for Unified Storage.

TCP connections

There is a possible cause for iSCSI timeout error messages occurring on client side. On the NAS boxes, it is considered that a max of 300 iSCSI sessions can be handled correctly. In iSCSI terminology, a 'session' is a single logical connection between an iSCSI initiator and an iSCSI target, and a 'connection' is a TCP connection. Each session can have multiple TCP connections for multipathing purposes (see http://utcc.utoronto.ca/~cks/space/blog/tech/ISCSIMCSAdvantages).
A common way to fall in a big amount of iSCSI sessions is to provide only 1 LUN per iqn target. If many LUNs are configured, this will increase the number of iSCSI sessions. To reduce this, just assign more LUNs to less iqn targets.

Number of iSCSI sessions per iqn target

Under some conditions it is possible to increase the number of iSCSI sessions per iSCSI target. On Solaris host, do as follows :


# iscsiadm list target-param -v iqn.1986-03.com.sun:02:173b14f2-2bce-ef3f-8e10-ba2d164b655d
# iscsiadm modify target-param -c 4 iqn.1986-03.com.sun:02:173b14f2-2bce-ef3f-8e10-ba2d164b655d
The max is 4 sessions par iSCSI target.

 

Back to <Document 1331769.1> Sun Storage 7000 Unified Storage System: How to Troubleshoot Performance Issues

See NAS wiki page for more info: https://stbeehive.oracle.com/teamcollab/wiki/AmberRoadSupport:iSCSI+is+hung+or+very+slow

References

PARTITION ALIGNMENT GUIDELINES: HTTPS://BLOGS.ORACLE.COM/DLUTZ/ENTRY/PARTITION_ALIGNMENT_GUIDELINES_FOR_UNIFIED
MULTIPLE CONNECTIONS/SESSION: HTTP://UTCC.UTORONTO.CA/~CKS/SPACE/BLOG/TECH/ISCSIMCSADVANTAGES
AMBER ROAD SUPPORT WIKI: SLOW ISCSI PERFORMANCE: HTTPS://STBEEHIVE.ORACLE.COM/TEAMCOLLAB/WIKI/AMBERROADSUPPORT:ISCSI+IS+HUNG+OR+VERY+SLOW
<NOTE:1331769.1> - Sun Storage 7000 Unified Storage System: How to Troubleshoot Performance Issues

Attachments
This solution has no attachment
  Copyright © 2012 Sun Microsystems, Inc.  All rights reserved.
 Feedback