I currently have a Seagate 1.5TB 7200.11 as my parity drive. I do have a WD 2TB EADS drive that I can install in place of the Seagate.

 

I've followed some threads where speed is combination of factors and some people noticed a huge increase in performance by using a 7200RPM Parity drive.

 

Sooo...several scenarios for me.

 

1) Replace my Seagate 1.5TB 7200.11 with the WD 2TB EADS drive for Parity and then move the Seagate 1.5TB to a data drive.

 

2) Replace my Seagate 1.5TB with some 2TB 7200RPM drive and move the Seagate 1.5TB to data.

 

3) Same as above but also install the WD 2TB EADS as a data drive.

 

Anyone have any personal experience with something like this? ie, Is it really that noticeable in day to day use? I've been quite happy with the current performance.

 

You MUST use the largest drive in your array as the parity drive.  If you don't, your array will fail as it reaches full capacity.  Having said that, you can pick up a Hitachi 7200 rpm 2TB drive for less than a 2TB Green will cost you.  I believe Weebo made a comment in another thread about the increase in parity writes by changing from a 5400 - 5900 rpm drive to a 7200rpm.  It was roughly a 40 - 50% increase if I remember correctly, which correlates to the spindle speed increase.

The speed of your parity drive will only affect your write speeds and your parity builds/checks.  Your read speeds will stay the same.  Also, adding a cache drive to your system will improve your write speeds far more than will using a faster (7200 rpm) parity drive.

 

So my advice is this (assuming you have a Pro license and use User Shares):

1) Remove the 1.5 TB drive and put the 2 TB drive in its place.

2) Start a parity build and wait for it to complete (your data will be unprotected during that time, but you'll still have the valid parity data sitting on the 1.5 TB drive as a backup if necessary)

3) Once the parity build completes, run a parity check

4) Once the parity check completes, power down the server and unplug (but do not physically remove) your smallest data drive

5) Plug your 1.5 TB data drive into the smallest data drive's slot and boot up the server.

6) Click 'I'm sure I want to do this' then 'Start' to rebuild the smallest data drive's data onto the 1.5 TB drive.

7) Once this completes, run another parity check (unnecessary, but I'm trying to give you all the safety checks I can)

Plug in your smallest data drive again and assign it to the 'cache drive' slot.

9) Edit each of your user shares and tell them to 'use cache drive'

10) Start the array and enjoy

 

This method will result in greatly increased 'perceived' write speeds to the server, since the cache drive is outside of the protected array (write speeds to my cache drive average 50 mb/s).  Your read speeds will be the same as they currently are.  You will also have more free space in your server to fill up with whatever you like (actual space being 1.5 TB minus the size of your smallest data drive).  Your parity build and parity check speeds will actually be a bit slower (since your parity drive is slower), but you should only have to build parity once and check it monthly, so it shouldn't really impact your day-to-day use of the server at all.  I currently use a WD Green drive (the older ones, not the EADS) as my parity drive and my parity checks still complete in less than 6 hours.

You MUST use the largest drive in your array as the parity drive.  If you don't, your array will fail as it reaches full capacity.

Actually, unless the assigned parity disk is as big, or bigger than the largest of your assigned data disks, the array will not start and just give you an error message stating the the parity drive is not big enough.

I believe Weebo made a comment in another thread about the increase in parity writes by changing from a 5400 - 5900 rpm drive to a 7200rpm.  It was roughly a 40 - 50% increase if I remember correctly, which correlates to the spindle speed increase.

This will ONLY be true if the data drives are also 7200 RPM.  Parity calc/check speed is limited by the rotational speed slowest of the drives involved.

 

The only time a faster parity drive will help is if the data drive it is being used with is also has as fast a spindle speed ... or, if multiple (slower) data disks are being written to simultaneously.  If you have a 2TB 5400 RPM data drive and a change from a 5400 RPM parity drive to a 2TB 7200 RPM parity drive it will not change the overall parity calc speed.

 

Joe L.

So, in all of my examples above the parity drive would be largest drive. I mentioned either using the 2TB WD EADS as the Parity drive and moving the Seagate 1.5TB to data or if the 2TB WD EADS was a problem I could just add a 7200 RPM drive and then add the 2TB WD EADS as a data drive.

 

I understand the comment about the cache drive but I haven't had the need for it as 25MB write speeds without a cache drive has been fine for me compared to my Drobo at work, that's about 2-3x faster .

 

I currently have all 7200RPM drives in my unRaid system. So it sounds like if I get a 2TB 7200RPM Parity drive and then install the 5400RPM WD drive, the only time it'll be slower is when I'm writing data to the WD drive whereas if I plug the 5400RPM WD drive in as parity since it has to do parity with all the other drives with a faster spindle it'll slow down anytime I write data anywhere.

 

currently have all 7200RPM drives in my unRaid system. So it sounds like if I get a 2TB 7200RPM Parity drive and then install the 5400RPM WD drive, the only time it'll be slower is when I'm writing data to the WD drive whereas if I plug the 5400RPM WD drive in as parity since it has to do parity with all the other drives with a faster spindle it'll slow down anytime I write data anywhere.

I think you have it exactly correct... with one tiny exception.  It might slow down your parity "check" speed to the speed possible reading the 5400RPM drive, but only if it cannot be read as fast as the 7200 RPM drives. 

 

Now, you might not notice any difference at all, because there may not be any difference when "reading" from the disks. When checking parity we are only reading the disks and have no need to wait for platter rotations between a "read" and a "write" of the same sector.  In other words, the raw data "read" speed of the slowest disk dictates the parity check speed (if buss or disk controller bandwidth does not limit you first)

 

Joe L.

I currently have all 7200RPM drives in my unRaid system. So it sounds like if I get a 2TB 7200RPM Parity drive and then install the 5400RPM WD drive, the only time it'll be slower is when I'm writing data to the WD drive whereas if I plug the 5400RPM WD drive in as parity since it has to do parity with all the other drives with a faster spindle it'll slow down anytime I write data anywhere.

 

This actually depends on how much data is being written. Consider for a moment that any write requires a read of the parity and data drive for the sector about to be written.

 

As mentioned multiple times, Write speed is dictated by the slowest drive in the path.

If a write is being issues to 1 drive, the parity and that drive is involved.

If the parity throughput can keep up with the data drive, then spindle speed would not be a bottleneck.

 

If there are two writes issued, the parity and the other two drives are involved. In this case, having a higher rotation on the parity drive helps. (i.e. torrents).

 

You can always do a test to see how the 5400 RPM drive affects you.

If your writes are sparse and small you may not perceive and difference.

Thanks everyone for their guidance. I'm going to think about it for a little bit. I have plenty of space at the moment and time to think .

FYI - I have the Hitachi 2TB 7200 drive as my parity drive.  My speeds are in the user benchmarks section.  I've benchmarked the drive as well as including the parity check speed and the preclear speeds to the tables. 

FYI - I have the Hitachi 2TB 7200 drive as my parity drive.  My speeds are in the user benchmarks section.  I've benchmarked the drive as well as including the parity check speed and the preclear speeds to the tables. 

 

Thanks! How do you like the Hitachi drive? I've heard that it runs a bit warmer than other drives? Does it run warmer than other manufacturer drives in your system?

FYI - I have the Hitachi 2TB 7200 drive as my parity drive.  My speeds are in the user benchmarks section.  I've benchmarked the drive as well as including the parity check speed and the preclear speeds to the tables. 

 

Thanks! How do you like the Hitachi drive? I've heard that it runs a bit warmer than other drives? Does it run warmer than other manufacturer drives in your system?

 

I have 2 of the hitachi drives in my system and they run great.  The temps are about in line with the others in my server, except the seagate drives which are a good 3-4C higher then the others.

My hitachi runs about 33* under load during parity checks.  My Seagates (2TB and 1.5TB LP drives) run about 30-31 under load.  My WDC Green 1TB drives run about 28-29*.  Those are the maxes I've seen on these so far.  This is in a Norco 4220 case with mechatronixs fans (about 30% slower (and quiter) than the delta screamers that case comes with).

Thanks! How do you like the Hitachi drive? I've heard that it runs a bit warmer than other drives? Does it run warmer than other manufacturer drives in your system?

I have no drives in my array to compare to, but without any cooling, my Hitachi 2TB's ran about 42 - 43C during a preclear cycle.  With a 140mm fan on a 4-3 drive cage the same drives run about 29 - 30C doing the same load.  They run about 125 - 130 MB/s on the edge of the platter, and about 60 MB/s on the inside.  They're also very quiet.

FYI - I have the Hitachi 2TB 7200 drive as my parity drive.  My speeds are in the user benchmarks section.  I've benchmarked the drive as well as including the parity check speed and the preclear speeds to the tables. 

 

Thanks! How do you like the Hitachi drive? I've heard that it runs a bit warmer than other drives? Does it run warmer than other manufacturer drives in your system?

 

I have 2 of the hitachi drives in my system and they run great.  The temps are about in line with the others in my server, except the seagate drives which are a good 3-4C higher then the others.

 

Can you tell me the model numbers of your other Seagates? Are they older models or newer ones? Trying to get a sense of whether they'd be warmer or cooler in my setup

Can you tell me the model numbers of your other Seagates? Are they older models or newer ones? Trying to get a sense of whether they'd be warmer or cooler in my setup

 

Oneis an old Seagate 750GB drive and the other is a 1TB drive.  The 750 runs the hottest when being used while the 1TB is about 2C lower.  I am pretty sure these are 7200.10 models.

Can you tell me the model numbers of your other Seagates? Are they older models or newer ones? Trying to get a sense of whether they'd be warmer or cooler in my setup

 

Oneis an old Seagate 750GB drive and the other is a 1TB drive.  The 750 runs the hottest when being used while the 1TB is about 2C lower.  I am pretty sure these are 7200.10 models.

 

Interesting...my Seagate 750GB is a Barracuda ES which definitely runs about 3-4C hotter than my Seagate 7200.11. It wouldn't surprise me if the Barracuda ES is the enterprise version of the 7200.10. Thanks!