Jump to content
  • 0

SSD Optimizer - Raid 0 Mechanical Drives


bunion

Question

Hi there, i'm new to Drivepool after Server 2008 Raid 5 has let me down with poor performance and difficulty identifying problem disks. The SSD Optimizer feature seems pretty cool - got a few questions on how I can make use of it with my setup if anyone can offer there guidance.

 

I don't have a spare SSD, what do you think of using two mechanical 2TB drives in Raid 0 as an SSD drive? Is this a waste of capacity?

 

Alternatively could I make a partition of my system SSD drive to use as the SSD landing, if so what is the optimum size?

 

Thanks for reading.

Link to comment
Share on other sites

13 answers to this question

Recommended Posts

  • 0

You can definitely use a RAID array for this instead of an SSD. We use "SSD" for the name because, well.... that's what most are probably going to use. 

 

And yes, you could use part of your system disk for this as well. But you will want two disks if you're enabling duplication, it will fall back onto another drive.

 

As for optimal size, that depends on usage. But basically, at least a bit larger than the largest size file that you plan on using.

More most, a couple of 60GB drives should be fine. However, if you plan on using larger files than that... well, the bigger the better, then.

 

 

As for the RAM Disks... that depends on how the software works. But unless it flushes the contents to disk on shutdown and re-adds them on startup.... chances are that it will cause issues.

 

Also, as I said, you will want to have TWO drives for this if you are using duplication. Otherwise, it will write to a HDD if it can't find enough "feeder/cache" disks.

Link to comment
Share on other sites

  • 0

For duplication, how much of a performance disadvantage would two 32Gb partitions on a 64Gb SSD drive be vs two SSD drives?

 

Edit: Sorry - just realized, is that a dumb question? Or can I still take advantage of the SSD landing on two partitions because it can be duplicated properly to other mechanical drives in the pool. Thanks!

Link to comment
Share on other sites

  • 0

Won't work. DrivePool is aware of volumes on the same disk, and won't use the same disk for duplication. This is because... well, if a disk fails, then you lose both copies. That's just stupid. And bad.

 

So this applies to the feeder disks as well, so you'd need two disks. If you have a large SSD for the system disk, you could use that and one additional SSD to get this working properly. Otherwise, one of the duplicates will fall back onto a mechanical/archive disk.

 

 

Also, if you're using greater than just 2x duplication, you'll need that many disks.

Link to comment
Share on other sites

  • 0

It's not like I would ever use it but given that I think I've seen a number of questions on feeder disks, I wonder whether it would make sense, even for a Pool with duplication, to specifically allow files not to be duplicated while on a feeder disk? The thinking is (and I'm not entirely sure if this is correct) that files would/should not reside on feeder disks for long as, when I/O is low, they'd be moved to normal drives. Assuming a user is willing to take a risk for a short while, could not he/she have just one copy saved to a (SSD) feeder disk and it only be duplicated when it's moved off of it?

Link to comment
Share on other sites

  • 0

Just to be clear, if you have 1 "SSD" drive and multiple "archive" drives, the following should happen?

  1. New writes to a duplicated folder should go to the SSD, then when drivepool next initiates a balancing of the pool, a copy of the new files will be transferred to 2 archive drives. The new files originally copied to the SSD will then be removed. OR are new writes simultaneously written to the SSD and 1 archive disk, then a second copy made to another archive disk and the SSD's copy deleted upon balancing?
  2. New writes to a non-duplicated folder should go to the SSD, then when drivepool next initiates a balancing of the pool, a copy of the new files will be transferred to 1 of the archive drives (priority given based on other balancer settings). The new files originally copied ot the SSD will then be removed.

I ask because I have a very simple test setup including 1 SSD and 2 Archive drives and while monitoring writes in both scanner and drivepool, the SSD is rarely, if ever, used (tested copying files to both duplicated and unduplicated folders). Sometimes the SSD will be used to copy the very 1st file in the transfer and nothing more. For the sake of identifying the issue i turned all other balancers off and have the following settings (see image). I've also tried turning off all of the "ordered placement" options. 

 

Am I misunderstanding this tool?

 

thanks!

 

SSD%20Optimizer.jpg <https://www.dropbox.com/s/d56zbf3x8g3nxm4/SSD%20Optimizer.jpg>

Link to comment
Share on other sites

  • 0

From the notes of the balancer:
 

  • If you are using duplicated files, then you should specify multiple SDD disks or else the system will fall back to an Archive disk for one or more of the file parts.

This means that if you have duplication enabled and write to a folder that is duplicated, it will write to two disks. That means that it may slow down the write if there is only one "fast" drive.

 

That is why we recommend two SSDs if you intend on using this balancer.

Link to comment
Share on other sites

  • 0

ya, I get that. The majority of my confusion actually comes from that same document, located here, which states, "SSD disks will receive all new files created on the pool.  It will be the balancer's job to move all the files from the SSD disks to the Archive disks in the background." -definitely not what i'm seeing.

 

(drashna- you are also implying that if a folder is flagged for duplication, any file copied to it will be instantly copied to 2 drives, outside of any balancing tasks or balancer rules, correct? ...DP wont wait for the next scheduled re-balance to duplicate)

 

also (UPDATE) I was seeing my SSD being completely ignored when transferring a folder of files (a few GB, multiple file sizes) to / from both duplicated and unduplicated folders in the pool.  However, I just got back from a trip, fired up the server, and things are working much more in line with the balancer's documentation. I had already tried a restart when i was troubleshooting so hmmm.... I guess as long as its working properly now, I shouldn't care why it was acting up before :) !

Link to comment
Share on other sites

  • 0

By default, the the drives marked as "SSD" should recieve the files first.

You many notice red/orange arrows that appear below the bar graphs for the drives.

These are the "real time placement limiters". All the archive disks should have them at the beginning of the drives, indicating that they should not be used to place files on immediately.

And the SSD drives should have them at about 70%, meaning that's how full they can get before using another disk.

 

If the pool was already balancing, it may not have shown them yet. 

But when you copy files, it should use the real time placement limiters for determining where they go first, and then follow any other enabled balancers.

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Answer this question...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
×
×
  • Create New...