Solid State Drives in RAID -> TRIM

[Malonn]

I have 2 OCZ Vertex 3's in a RAID 0. Everything I've read before says that TRIM is not supported when running a RAID array. But I checked in the console (command prompt, whatever), and it said TRIM is active and running. Plus I downloaded a SSD tweaking app, and it says TRIM is good to go. It even tweaked TRIM.

Is this an anomaly? Or is TRIM really not working and it is being reported that it is supported by hardware, but it isn't actually doing anything? I need to know because I'll need to download an app to maintain my drives if TRIM isn't working.

[lowfat]

There is no way to see if the TRIM command is getting passed to the drive. All you can check is to see if the OS is sending the TRIM command.

At the moment there is no way to get TRIM w/ RAID0. No matter what RAID controller you are using.

[Ian[Frozen]]

Yes what he said

At the same time who cares as it's not needed

[Malonn]

It's not? Well, just for good measure, what are my alternatives for the same result as TRIM? I tried Wiper.exe, but it wouldn't work.

[lowfat]

You just have to rely on ITGC (idle time garbage collection). Most modern SSDs have good ITGC, however it is a little harder on NAND than if it were to just get the TRIM command AFAIK.

[Ian[Frozen]]

Wiper is an Indilinx utility and does not work on a Sanforce based drive, let's hope no harm done...

Meanwhile, have been using GC in one form or another for at least 2years. Obviously without Trim these drives are performing almost as good as new, with improvements to raid controllers these drives + latest built-in GC actually perform better now than any results they achieved when new. It has been proven that GC can at least maintain performance which Trim alone can not accomplish.

Trim IMO is already obsolete so why would any manufacturer pursue further develop?

[Ao1]

It has been proven that GC can at least maintain performance which Trim alone can not accomplish.

Link?

SSD's with really aggressive GC wear out NAND much faster than a drive with less aggressive GC. Newer drives have less aggressive GC due to the reduced write capability of higher density NAND.

If you need TRIM or not depends on your work load.

Most desktop users are only going to use a tiny fraction of the IOP capability of a SSD so you will not notice if the drive is degraded in normal use.

From Anandtech......

General comment on drives without TRIM:

"You'll notice a huge increase in max latency compared to average latency, that's because this is when a lot of drives do some real-time garbage collection. If you don't periodically clean up your writes you'll end up increasing max latency significantly."

Intel G3

"The 320 behaves a lot like the old X25-M G2 did when tortured. Minimum performance drops pretty low - Intel prefers cleaning up as late as possible to extend drive longevity. As a result, I wouldn't recommend using the 320 in an OS without TRIM support."

C300/ M4
"Ugh. This graph looks a lot like what we saw with the C300. Without TRIM the m4 can degrade to a very, very low performance state. Windows 7's Resource Monitor even reported instantaneous write speeds as low as 2MB/s. The good news is the performance curve trends upward: the m4 is trying to clean up its performance. Write sequentially to the drive and its performance should start to recover. The bad news is that Crucial appears to be putting off this garbage collection work a bit too late."

Kingston SSDNow V+ 100
AFAIK the Kingston SSDNow V+ 100 is the only SATA based SSD that does not need TRIM to maintain performance. Again from Anandtech...

"The benefit of this [aggresive GC] is you get peak performance out of the drive regardless of how much you use it, which is perfect for an OS without TRIM support - ahem, OS X. Now you can see why Apple chose this controller.

There is a downside however: write amplification. For every 4KB we randomly write to a location on the drive, the actual amount of data written is much, much greater."

Sandforce

[Gilgamesh]

If I create a raid sw (striping) with Windows 7 Disk Management on 2x ssd (no boot device), Trim command is passed on drives?

[Ian[Frozen]]

That's all well and good, however where you see Trim fail is over time ie: several months of usage. Read the forums, many folks relying on firmware with Trim only has eventually suffered sever performance degradation. This fact was I believe the driving force to include GC within the firmware which will function so long as the drive is powered. All Vertex drives since firmware 1.41 have included GC, the latest revisions of GC are less aggressive. I thin GC has been in use since mid 2009 and as yet TMK no signs of any drive wearing out.

RAID users are reliant on GC to maintain their array in peak performance. The longest practical GC test I've run on my array is 6 months (with used nand) which indicated no degradation of performance. AFAIK Trim can not make the same claim, it certainly has not been my experience which has shown single drive deteriorate within 2 months.

[Ian[Frozen]]

If I create a raid sw (striping) with Windows 7 Disk Management on 2x ssd (no boot device), Trim command is passed on drives?

No, there are currently no raid controller drivers capable of passing the command to a drive.

RAID users rely totally on GC to maintain their drives.

[Ao1]

If I create a raid sw (striping) with Windows 7 Disk Management on 2x ssd (no boot device), Trim command is passed on drives?

I believe Anvil established that TRIM does work with software raid.

EDIT
Here is Anvil's post that proves TRIM works in SW raid
http://www.xtremesystems.org/forums/...91#post4626691

Although the TRIM command captured by hIOmon does not in this instance confirm that TRIM was being executed past the volume level that was being monitored, the delete operation would appear to indicate that a TRIM operation was triggered enabling performance to be restored.

[Ao1]

Read the forums, many folks relying on firmware with Trim only has eventually suffered sever performance degradation.

Maybe you see that on OCZ forums, but that I believe that is due to wear level throttling kicking in, not a failure of TRIM.

[Gilgamesh]

I believe Anvil established that TRIM does work with software raid.

EDIT
Here is Anvil's post that proves TRIM works in SW raid
http://www.xtremesystems.org/forums/...91#post4626691

Although the TRIM command captured by hIOmon does not in this instance confirm that TRIM was being executed past the volume level that was being monitored, the delete operation would appear to indicate that a TRIM operation was triggered enabling performance to be restored.

Thank for reply

[Ao1]

I tried sw raid and from what I can recollect it was not as good as onchip raid performance. I think F@32 had a thread going at some time on it. It would be worth looking it up.

I don't know if Anvil did any further testing to verify TRIM functionality. I'm sure he will chip in if he has.

[Ian[Frozen]]

This debate really heated up some time back when Intel published their latest driver update notes by misleading us into believing Trim commands were passed by the driver in raid mode. Whereas the true meaning clearly indicates that Trim commands are passed only under two scenarios. first on single drives configured as non raid mode, secondly (and more importantly) are drives configured as NON-MEMBER drives in RAID mode whether or not there is a raid array present. This means that drives as part of a raid arry do not receive trim commands..

Additionally, AFAIK there are no raid card manufactures with current drivers which support trim to date.

If raided drives had at any time received a trim command, why is it that raided drives pre-GC suffer such performance degradation??