performance questions

Software-based VM-centric and flash-friendly VM storage + free version

Moderators: anton (staff), art (staff), Max (staff), Anatoly (staff)

User avatar
anton (staff)
Site Admin
Posts: 4010
Joined: Fri Jun 18, 2004 12:03 am
Location: British Virgin Islands
Contact:

Thu Jan 25, 2007 10:39 pm

You need to do performance tuning with RAM disk. Untill you did not get WIRE SPEED with RAM disk - your network configuration is not optimal. This is ONE.

How fast is hard disk itself? Use HDTach to benchmark it locally and also use HDTach to benchmark it with ImageFile created (SPTI SHOULD be slower as it's sync and cache disabled) over the very beginning of the 1st partition. Difference should be minimal. This is TWO.

Please report us 1) and 2) statuses :)

Thanks!
yyrkoon wrote:Well, anton, I'm not sure what to say, the Intel pro 1000 PT card (PCI-E) did make a difference, but not a whole lot. This time, I used Starwind + Starport, and the best transfer I recieved was an average of around 40MB/s, with a peak of around 55MB/s, using an img file. SPTI is much slower (between 20-30MB/s )

This is what I've done, just in case I've missed something important:

1) Installed Starwind on Target machine, enabled jumbo frames, and checked all other hardware options for the adapter.
2)Install Starport on initiator, enabled 9014 byte Jumbo frames, checked, and changed any necessary hardware options.
3) Followed instructions here -> http://www.starwindsoftware.com/forums/ ... mbo+frames, but instead of putting it in the key you suggested, I did it for the individual adapters, in their respective locations
4) Setup the disks on the Target.
5) Initialized the 'disks' on the Initiator side, and formated with varying block sizes, which didnt seem to make much, if any difference(at least towards better speed, sometimes it did get worse).

Anyhow, I'm almost hoping I missed something somewhere, because ~30MB/s is what I'd also get using Samba, or other file level access method, and we all know, iSCSI HAS to be better than Samba. It is really frustrating. If the above is not the case, then perhaps my hardware is holding me back, but to be honest, it is not worth spending large amounts on new hardware to find this out (at least, not for me).
Regards,
Anton Kolomyeytsev

Chief Technology Officer & Chief Architect, StarWind Software

Image
yyrkoon
Posts: 8
Joined: Sun Dec 10, 2006 10:08 pm

Fri Jan 26, 2007 10:36 am

Ok, will give it a try, in a couple of days, I've done SO MUCH testing over the last several months, that I need a break, even if I was having fun . . .

[EDIT]

Oh, on a side note, I setup Windows2003 server on a friends PC this evening, which has newer equipment in it, using a Promise sx4060 HBA, and 3x Seagate Barracudas 250GB drives in RAID0 (4th one died prematurely in RAID5, RMA time . .), motherboard is an ABIT NF-M2 nView, and I do realize it is not server grade, but I gave it a quick test with Starwind, anyhow.

Just glancing at Task Manager / Network,, performance, drag, and drop directory of ~4GB, mainly a large ISO, and a few binaries of ~20MB, I saw bytes sent peak at around ~60,000,000 bytes/s, and writing back, peaked at ~80,000,000 bytes/s. These are the highest readings I've seen yet using an img file, but it's not consistent, and yes, I do realize that PerfMon isn't all together reliable either.
User avatar
anton (staff)
Site Admin
Posts: 4010
Joined: Fri Jun 18, 2004 12:03 am
Location: British Virgin Islands
Contact:

Fri Jan 26, 2007 9:06 pm

What you do is a little bit unprofessional :) You're trying to "tune the radio with more then 1 handle at the same time". So you cannot isolate the issue and find out what's wrong. What you really have to do:

1) Get proper hardware. Even not server board with direct chipset embedded NIC (like the one on nearly 3 year old i875) would be good enough for full blown GbE wire speed. PCIe cards are a bit clumsy. PCI-X ones are EXCELLENT. If you'd get wrong hardware - bus speed would be limiting factor.

2) Pick up correct software settings, drivers etc. Use NTtcp and IPerf to measure NIC performance. You should get 900-950 megabit. If you'd skip this step - impoperly configured TCP stack would be limited factor (In our test machines we usually get 400-600 megabit BY DEFAULT).

3) Configure iSCSI target to use RAM disk and make sure you get wire speed with iSCSI as well. I/O meter, HD tach and true file copy would be in help here.

4) Play with RAIDs making sure they are fast enough (same test from p.3 run LOCALLY).

That's RIGHT metodology. Everything else is WRONG. More or less :)
yyrkoon wrote:Ok, will give it a try, in a couple of days, I've done SO MUCH testing over the last several months, that I need a break, even if I was having fun . . .

[EDIT]

Oh, on a side note, I setup Windows2003 server on a friends PC this evening, which has newer equipment in it, using a Promise sx4060 HBA, and 3x Seagate Barracudas 250GB drives in RAID0 (4th one died prematurely in RAID5, RMA time . .), motherboard is an ABIT NF-M2 nView, and I do realize it is not server grade, but I gave it a quick test with Starwind, anyhow.

Just glancing at Task Manager / Network,, performance, drag, and drop directory of ~4GB, mainly a large ISO, and a few binaries of ~20MB, I saw bytes sent peak at around ~60,000,000 bytes/s, and writing back, peaked at ~80,000,000 bytes/s. These are the highest readings I've seen yet using an img file, but it's not consistent, and yes, I do realize that PerfMon isn't all together reliable either.
Regards,
Anton Kolomyeytsev

Chief Technology Officer & Chief Architect, StarWind Software

Image
Post Reply