1TB 2.5@7200 vs 2TB 2.5@5400


Recommended Posts

A question for you guys who know about areal density and how it effects speed.

 

Need a couple of cache drives.  1TB 7200RPM disk is $69, 2TB 5400RPM is $79. 

 

Writing to the array is my only concern, speed is key, I'm not sure I would ever actually fill a 2TB cache disk before the mover gets a chance to run.  Still, its 2TB.

 

Which is likely to be the faster drive for writes?  I assume each has only one platter (I need to research this more, but for purely scholarly discussion let's assume one platter in each disk)

 

If it matters, I am comparing an HGST H2IK10003272SN to a Samsung Spinpoint M9T.

 

Thanks.

Link to comment

Laptop drives top out at 667GB/platter areal density these days.

 

I can't find the specifics on the HGST drive, but I suspect it's a 500GB/platter unit.

The Samsung M9T is a 667GB/platter unit.

 

So the sustained data rates will be virtually identical, since a 500GB platter spinning at 7200rpm will almost exactly match the throughput of a 667GB platter unit spinning at 5400rpm.

 

But the seek time will be appreciably better on the faster drive -- so your overall performance will be better with the 7200rpm unit.

 

Of course if you REALLY want to emphasis performance, you may want to consider a 500GB SSD ... these are available for under $200 these days  :)

 

 

Link to comment

The 2TB drive will deliver better sequential performance, as it has 3 platters. But beyond that special case, the mass of the heads slows the drive down. Both report 12ms seek times and 32MB cache.

 

How big a cache drive do you need? Spending $10 for space never used is not worth it.

Link to comment

Thanks to all.  Definitely NOT sold on the SSD idea yet.  For a cache drive, with my primary desktop using the secondary unraid server as primary storage, and with zliions of writes possible from regular use, I still freak out when I think about SSDs dying early deaths because they get worn out.  I'll use an SSD as the OS drive in my next build, but never for data.  At least not yet.

Link to comment

I can understand that ... although I don't think modern SSD endurance levels (typically 500TB of writes) are really a problem.    However, it's also true that any good rotating platter drive can easily sustain writes at speeds fast enough to saturate a Gb network ... so when used as a cache there's little (if any) advantage to an SSD (the initial start of a write is faster due to seek time, but that's likely masked by UnRAID's buffering).

 

 

Link to comment

... Definitely NOT sold on the SSD idea yet.  For a cache drive, with my primary desktop using the secondary unraid server as primary storage, and with zliions of writes possible from regular use, I still freak out when I think about SSDs dying early deaths because they get worn out.  ...

I wouldn't worry about it. I can't remember the exact numbers and don't have the link now, but when I was researching use of SSD for intensively used cache drive. I came across third-party "exhaustion" tests data for 1TB Samsung 840 EVO and the numbers were such that if I write 100 GB every day - this drive will easily last for at least 12 years.

Link to comment

... Definitely NOT sold on the SSD idea yet.  For a cache drive, with my primary desktop using the secondary unraid server as primary storage, and with zliions of writes possible from regular use, I still freak out when I think about SSDs dying early deaths because they get worn out.  ...

I wouldn't worry about it. I can't remember the exact numbers and don't have the link now, but when I was researching use of SSD for intensively used cache drive. I came across third-party "exhaustion" tests data for 1TB Samsung 840 EVO and the numbers were such that if I write 100 GB every day - this drive will easily last for at least 12 years.

 

I have over a dozen of the Samsung EVO SSD's and I have not had one failure yet.

In fact, I was really surprised this weekend.

I was doing some cloning from magnetic drive to ssd, I pulled out my storage box of 2.5" s and realized I had over a dozen SSD's.

None have failed yet.

 

I believe the early studies said you can write 20GB a day and they will last for 5 years.

Keep in mind you need to keep some cells free.  Therefore I suggest over provisioning the drive to the size you think you need.

 

I use my laptop 24x7x365 with capturing music on an SSD. It's writing constantly.

The drive is over 3 years old and is still showing healthy.

Link to comment

...

I use my laptop 24x7x365 with capturing music on an SSD. It's writing constantly.

The drive is over 3 years old and is still showing healthy.

 

Could you post an updated SMART status of this drive?  :P

 

I'm very interest on this since I've one server on production recently updated to SSD... (3x1TB 840 EVO in RAID5)

Link to comment

...

I use my laptop 24x7x365 with capturing music on an SSD. It's writing constantly.

The drive is over 3 years old and is still showing healthy.

 

Could you post an updated SMART status of this drive?  :P

 

I'm very interest on this since I've one server on production recently updated to SSD... (3x1TB 840 EVO in RAID5)

 

 

The Samsung SSD Magician does not export the SMART values in an easily capture-able manner.

Link to comment

Thanx a LOT!!!  :D

 

From a fast analisys of data, the second ssd is still perfect (newer!), while the first one (840 evo) is showing first sign of age but it still has a lot to live!  ;D

 

It in fact has already 8 blocks gone and a decreasing value of wear leveling...

 

Tomorrow I'll do some math about days of usage and total GB/TB already written to it.

 

What's the size of the 840 evo? (I need for my math...  :P)

Link to comment

Thanx a LOT!!!  :D

 

From a fast analisys of data, the second ssd is still perfect (newer!), while the first one (840 evo) is showing first sign of age but it still has a lot to live!  ;D

 

It in fact has already 8 blocks gone and a decreasing value of wear leveling...

 

Tomorrow I'll do some math about days of usage and total GB/TB already written to it.

 

What's the size of the 840 evo? (I need for my math...  :P )

 

It's a 250GB Samsung 840 PRO.

It was purchased and put in use sometime around December of 2012

 

Both the Samsung magician and SSDlife say it's healthy.

Link to comment

Both the Samsung magician and SSDlife say it's healthy.

 

Well, more serious software like HD Sentinel, expecially If set in "server strictly mode", will surely report it not totally healthy (you can try it, trial is free...), but, anyway, it still has tons of spare reserved blocks to use! And could still work fine for years...

 

Only be warned that when first bad blocks come out, sometimes, some SSD could die very rapidly after that... so, If you have valuable data on it, backup, backup, backup... ;)

 

Some maths will follow... :)

Link to comment

Both the Samsung magician and SSDlife say it's healthy.

 

Well, more serious software like HD Sentinel, expecially If set in "server strictly mode", will surely report it not totally healthy (you can try it, trial is free...), but, anyway, it still has tons of spare reserved blocks to use! And could still work fine for years...

 

Only be warned that when first bad blocks come out, sometimes, some SSD could die very rapidly after that... so, If you have valuable data on it, backup, backup, backup... ;)

 

Some maths will follow... :)

 

It's been in service for over 800 days. I beat on this thing 24x7x365.

I hardly ever turn the laptop off, only when I travel with it.

It's actually outlived my expectations.

I had purchased a 500GB to replace it, but still have not found the need for the extra space since I use my unRAID server for the bulk of my data repository.

 

I'll image it to another spare SSD to be on the safe side.

Sheesh I have a box of them thinking I always needed spares for the critical situation.

Newegg always has these sales.. and I'm such a sucker.. LOL.

I only wish they had a good sale on the Samsung EVO mSATA modules.

So far my Samsung evo's have performed way past my expectations.

Link to comment

... Newegg always has these sales.. and I'm such a sucker.. LOL.

 

You're not alone  :)

 

Don't have a dozen ... but do have 4 unopened SSDs between 240GB & 500GB

 

I DO plan to use them in the not-too-distant future; although I've already decided my next main PC will use a TB SSD, so I'll have to buy another one  :) :)

Link to comment

... and I also have an 8TB Seagate Archive drive coming this week => $230 was too good a price to pass up [Absolutely no currently projected use for it ... I'll just stick it in my desktop for now]

 

 

I missed that one. I've been meaning to acquire one for backups.

 

It wasn't a Newegg deal ... it was at NCIX.  The drive was $240, and since I'm a new customer there I got a $10 additional discount  :)    Just noticed they're now $332.98 and on back order, so it was indeed a very good deal.

 

I'm almost wishing I'd bought a half dozen and just built a new server  :)

 

BTW, as I assume you know, these drives supposedly retail for $260, but the demand has so far kept them above that for the most part.    Hopefully by the time I'm really ready to build another server with 6-8 of them Newegg will have a sale in the $250 range  :)

Link to comment

...

Some maths will follow... :)

 

No joy... Weebo your LBAs written RAW value is inconsistent with mine... if I got it as good, you should have only written about 700 GB to the drive, that is obviously not possible...  ???

 

Have you past 99999999999999 and come back to 0?  ;D

Link to comment

HD Sentinel says 14.76 TB, 818 days.

 

According to SSDlife, it says I wrote 94 GB in the last 7 days, 41 GB yesterday.

 

This seems about right, I am capturing mp3 music all day and all night.

Edit tags during the day and move it to the unRAID server.

 

41GB seems like allot but this is also my main workstation and I'm on it all day long.

Link to comment

HD Sentinel says 14.76 TB, 818 days.

...

 

Perfect. HD Sentinel is very sharp on 'intepreting' different methods used for RAW values, so this is likey the right value...

 

About 15 TB on a 250 GB drive gives back a 60 writings/capacity factor that is not so high after all...

 

BUT, since the use you described, I think that every single cell has been written A LOT more because of a lot of small writings... so your wear level is A LOT higher than it should be with same writings/capacity on different usage...  ???

(this without considering trim, garbage collection and wear leveling mechanisms that increase the number of real rewritings...).

 

On my production server every single 840 evo (RAID5) has got about 2,5TB of writings in about three months of use. Some has gone with first setup, some for a BAD bug with an antivirus (discovered and corrected after some days of crazy writings... :-X), so I think I'm going about with half a TB/month.

My drives are 1TB in size, so to go to a factor of 60 as you (where/when, with that SMART values I would change them, since of the 'mission critical' tasks performed by that server), it will require to me about 10 years of this usage to need them to be changed!

 

Surely more than their real usage before changing them with newer and faster hardware. Nice.  ;)

 

For hard usage is ever better to go with high capacities and RAID0/5 to obtain a longer lifetime and better performance so it seems I've well done my homeworks...  :D

Link to comment

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  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.