Staggered spinup on WD GP drives on cold start

XS Janus · Post by **XS Janus** » Tue May 06, 2008 11:43 am

I have a 3ware 9650SE controller and 4 wd1000fyps drives.
I'm trying to make this staggered spinup thing work when cold starting the system. But no matter how I connected them on my 3ware controller the drives startup all at once.

The thing is, in windows, after the drives fall asleep and you activate them, they will do staggered spinup just as it was set on the controller, just not on boot up

I did the tests again and I guess I noticed wrongly before.
The GP drive do power immediately on when power is connected to them.

I've set up the 3ware controller to spin up 1 drive at intervals of 1sec. and put SATA mode to SATAOOB
All drives have PM2 enabled via jumper on the drives.

I did notice that, "sometimes" and I do stress sometimes, while in enclosures, on cold start the drives in the same cage do start up in 2 instances (maybe 3 drives and a while latter 1 more ??)
Could something override controller settings and sometimes decide my current PSU has sufficient juice?

These are the things I've tried:
1. 3xGP drives powered by 4->15pin power connectors direct attached on 3ware via sata ML cable
2. 3xGP drives powered by PSU provided SATA power connectors direct attached on 3ware via sata ML cable
3. 3xGP drives powered by 4->15pin power connectors inserted in a straight through SATA ICYDOCK back-plane drive enclosure attached on 3ware via sata ML cable
4. 3xGP drives powered by PSU provided SATA power connectors inserted in a straight through SATA ICYDOCK back-plane drive enclosure attached on 3ware via sata ML cable

The result is always the same: the drives power-up simultaneously, and while in windows, after they fall asleep they will obey 3ware and spin-up one at the time.

Now I saw on the web that insulating power pin11 could make the drives wait for power-up command from controller, BUT i also saw that pin11 is used for drive activity monitoring and THAT function works very well on these enclosures and I would hate to loose it by insulating Pin11.

Anyway i don't think I could insulate pin on the enclosure itself, its out of reach. Maybe it could be done on the hdd itself. But I can't believe all this is required just to enable this feature...

WD info on pin11, look under LED:
http://www.wdc.com/en/library/sata/2579-001097.pdf

I also found this:
http://3ware.com/KB/article.aspx?id=14889 The same jumper settings already posted, and
http://3ware.com/KB/article.aspx?id=12072, which mentions some utility but I haven't noticed it doing anything to the drives.

Any ideas?

If someone had a similar occurrence please tell.
That would be most helpful and very much appreciated.

Sorry for the long post, and thanks once again for any input anyone can provide.

sjoukew · Post by **sjoukew** » Tue May 06, 2008 12:06 pm

I don't have any clue about the problem or the solution but there is one thing which you can change which isn't 100% correct if you ask me.
The spin up time from 1 hard disk is at least 10 seconds, as long as it takes to bring the drives up to full rpm. The huge power draw from the spin up continues for a couple of seconds, 10 or so. If you don't want the power draw from all disks spinning up all together, you should set the interval to at least 10 seconds. Else it doesn't help.

dhanson865 · Post by **dhanson865** » Tue May 06, 2008 6:00 pm

sjoukew wrote:I don't have any clue about the problem or the solution but there is one thing which you can change which isn't 100% correct if you ask me.
The spin up time from 1 hard disk is at least 10 seconds, as long as it takes to bring the drives up to full rpm. The huge power draw from the spin up continues for a couple of seconds, 10 or so. If you don't want the power draw from all disks spinning up all together, you should set the interval to at least 10 seconds. Else it doesn't help.

That's a bit outdated. Modern drives don't take that long to spin up. 1 second would be enough to keep the peak from being as high though not by much. Looking at the graph 2 to 5 seconds would be a reasonable range.

http://www.storagereview.com/guide/spinPower.html

Besides the Green Power drives are very new and draw low amounts of power. They are probably less of an issue than your average drive on a RAID array.

There are some drives out now that can spin up fully in less than one second. I have no idea if the Green Power series is in that class.

But you may be right on one thing, the OP might have been expecting less overlap and upping the delay might give him results closer to what he expects to see.

dhanson865 · Post by **dhanson865** » Tue May 06, 2008 6:16 pm

GreenPower drives consume less current during start up allowing more drives to spin up simultaneously resulting in faster system readiness.

Maybe the GP drives ignore spinup delay requests by design? Or the controller itself overides the delay if it isn't necessary?

I'm shooting in the dark here but if your not seeing reboots or data loss is it really an issue?

XS Janus · Post by **XS Janus** » Wed May 07, 2008 6:22 am

Well I plan on hooking it up to a Pico-PSU for low power. So I may run into the problem.
Cool graph, BTW!

Anyway, I know that it should work in staggered mode in boot up as well. My system disk is on the mobo controller, only the array is on 3ware.

The question is why won't it work that way: sata power connectors not supporting it, controller "sees" that my current psu won't have a problem with starting them all at once, or maybe some thing else?

Has anyone ever experimented with this kind of spinup mode? Any special settings you did to your hardware to make it work from boot up?

dhanson865 · Post by **dhanson865** » Wed May 07, 2008 7:32 am

You do have the drives per spin up set at 1 don't you? As seen on page 84 of this manual (note page 84 in the manual is page 94 in the viewer)

Oh and I noticed that the maximum spinup delay on that controller is 6 seconds. I can't imagine wanting to set it higher than 4 seconds with a really modern drive.

XS Janus · Post by **XS Janus** » Wed May 07, 2008 12:21 pm

Yes I've setup it just like that, tried other variations too.
longer periods don't do squat.

I'll try hooking up a 10$ psu to it tommorow and see if something moves.

sjoukew · Post by **sjoukew** » Sat May 10, 2008 4:22 am

@dhanson865

Thanks for updating my knowledge

I tried it indeed with an old drive in an external hdd cage. Good to see improvements in this area also

It is a nice graph

dhanson865 · Post by **dhanson865** » Wed May 21, 2008 7:38 am

Doing some further research on platter sizes today and accidentally found more power graphs.

http://www.seagate.com/support/disc/man ... 02371a.pdf (page 21 and 22)

On the Seagate 7200.10 series they don't pull much of any power until 2.5 secs in and then power stays high until about 5 secs, then tapers down until about 9.5 seconds.

7200.10 series of drives only guarantees spin up in under 15 seconds.

7200.7 series guarantees spin up in under 10 seconds.

On that series heavy power draw starts at about 1 sec in and continues until the 6 sec mark without dropping gradually like the newer drive does.

http://www.seagate.com/support/disc/man ... 7200pm.pdf (page 9)

Apparently spin up times vary quite a bit and there is no correlation between spin up time and how long the high power draw lasts.

On the GreenPower RAID edition drives we have http://www.wdc.com/en/library/sata/2879-701236.pdf

Drive ready time (2-disk) 11.0 sec (average)
Drive ready time (3- and 4-disk) 17.0 sec (average)

Which is longer than I expected. I haven't seen power graphs for WD drives and the Seagate drives show us that the power draw graph can vary in shape drastically so it's anyones guess how long you should set that delay for.