Bad performance with BlockIO

[Raudi]

Something i noticed... Can it possible that in BlockIO the local memory will not used for caching? If i remember right, a frew weeks ago i was wondering about much free memory on my system. Now all memory is used for system cache.

Stefan

[To-M]

Did you test with the Demo-CD - DSS yet? Also send your results to support with new logs.
In blockIO we use only cache specified for devices - fileIO use filesystem cache and devices cache. The default maximum of device cache in 32bit kernel is limited actually to about 1GB.

This was emailed to me today asking them to perform some test to verify there speed with iSCSI-R3 with 4x1Gb - MPIO from Microsoft. There must be something that needs to be tweaked on the hardware (firmware, NIC driver....), they testing with the Demo-CD then send the logs using the same ticket number.

hardware specification

Target

iSCSI-R3 Ent. Ver. 2.30 build 2820
CPU 2 x QUAD (Xeon) 2Ghz
Ram: 2Gb
Controller: LSI Logic SAS
Disc: 8x36GB (Raid0)
NIC: Intel Corporation Pro/1000 PT Quad Port LP server Adapter
Information: SSR212mc2

Client

Windows 2003 server with service pack 1
CPU: Xeon 1.86 Ghz
Ram: 2Gb
NIC: Intel Corporation Pro/1000 Pt Quat Port LP server Adapter
initiator: Windows iSCSI Initiator 2.05 build 2392


Performance Result

IOmeter MPIO , out_io=100 row-device
1worker 64k seq write io=3963,72 mb=247.73 cpu=70.78 1worker 64k seq read
io=5951,91 mb=371,99 cpu=81,76

2workers 64k seq write io=3776,17 mb=236.01 cpu=76,43 2workers 64k seq read
io=5368,14 mb=335,51 cpu=75,64

1worker 512 seq write io=700,37 mb=0.34 cpu=2,35 1worker 512 seq read
io=9264,68 mb=4,52 cpu=21,42

2workers 512 seq write io=1478.56 mb=0.72 cpu=5,76 2workers 512 seq read
io=13180,47 mb=6,44 cpu=22,88

[Raudi]

First my question: Is it possible to create a BlockIO target in a unit whre are already existing 2 FileIO targets? I don't want delete my data again only for a test.

And i tryed to download the Demo-CD but the link doesn't work. I logged in with my serial, selected the Demo-CD and then i got a error Message that the page cannot be displayed...

Greetings
Stefan

[To-M]

You cannot select a specified target to be BlockIO. As stated in release notes it is recommended to back up your data and remove volumes and recreate them.

Thanks for the link issue - you are correct there is a problem we will correct it.

Email me todd.maxwell@open-e.com and I will provide another site as this is not to be published on the Forum.

[Raudi]

The release notes told nothing about mixed enviroments, they told only about volumes and not about volume groups or units.

At this moment my default iSCSI target type is FileIO, i have free space in my volume group. If i understand right i can now switch to BlockIO on the console and create a new volume in the free space of the volume group. The result will be that i have my 2 old iSCSI volumes in FileIO and a new one in BlockIO.

Correct?

I got a demo CD, i used my e-mail address to get a access code, with this code i can download the CD.

Greetings
Stefan

[To-M]

New volumes are created in Block IO and as stated below in order to use old volumes that are created in File IO the data will need to backed up then restored on new volumes. You can still use your existing volumes as they where created in File IO but you will not be able to mix for Replication. In the end you would be better to keep them either File IO or Block IO.

Release Notes:

NEW:
* IMPORTANT : In ver 1.30 and 1.32 New iSCSI, default volume creation is done in block-IO in contrast to the older version that was file-IO.
In order to create new iSCSI volume in "old" file-IO mode, please switch the default in console tools menu: ctrl-alt-w --> Tuning options-->iSCSI daemon options-->iSCSI volume type.
Block-IO mode is about 30% faster then File-IO and the target volume size is exactly equal to defined iSCSI volume size (in file-io the target size is a bit smaller then defined iSCSI volume size).
Additionally, initialization is no-longer required as with the old file-io volumes.
In order to migrate your data from the file-IO to block-IO volume, the data must be backed up from the existing iSCSI volumes and then restored into newly created iSCSI volumes.
NOTE: please verify data integrity prior to deleting old iSCSI volumes.
Volume replication is only possible between to similar volumes (e.g. block-io <-> block-io or file-io <-> file-io). Volume replication between old file-io and block-io volumes are NOT possible.

[Raudi]

Ysterday i got the 2.32 and the small update for the intel drivers from the support to test it.

The problem is still the same. I still think it is a cache problem.

What i'm thinking:

With BlockIO there will be used a small cache so the read and write access is not optimized. With sequential read or write with one client or IOmeter with one worker there is no problem. But when i use more workers, i access the RAID with 2 clients at the same time or i made random access the performance goes extemely down.

With FileIO there will be used a large cache size (3,8 GB) so the writes can better ordered before writing to disk. So there will be more time for reading.

I made a test with IOmeter, this profile i used: smalltest.zip

Open-E S-ATA-Storage with FileIO:
64KB Random Read/Write Mix: 120 MB/s (80 MB read/40 MB write)
4KB Random Read/Write Mix: 15130 IO/s
1MB Sequential Read: 106 MB/s
1MB Sequential Write: 111 MB/s

Open-E S-ATA-Storage with BlockIO:
64KB Random Read/Write Mix: 14 MB/s (9,5 MB read/4,5 MB write)
4KB Random Read/Write Mix: 350 IO/s
1MB Sequential Read: 93 MB/s
1MB Sequential Write: 110 MB/s

My single local S-ATA disk is in random IO faster then the iSCSI storage in BlockIO...

Greetings
Stefan