Take the 2-minute tour ×
Server Fault is a question and answer site for system and network administrators. It's 100% free, no registration required.

Nowadays many server systems use 2.5" hard disks, afaik because they can put more spindles in the same chassis and maybe they use less power. When I don't need / cannot put more disks, is there any advantage in using 2.5" disks?

As an example, I can buy an 1U box with 4x3.5" or 4x2.5" disks for similar prices, but the 3.5" ones will be bigger (as in gigabytes, I mean). Of course I'm talking about similar disks (SAS 10k or SAS 15k, for example).

Given the same rotational speed and number of disks, do 2.5" disks have any advantage over 3.5" ones?

share|improve this question
    
YOu can also get a 1u case with 10 (!) 2.5" slots, you know. SuperMicro has them, dell, too. –  TomTom Dec 13 '11 at 12:38
    
You can get a 1U case with 12 2,5" slots if you need them to be accessible from the front. Otherwise, much higher densities are possible - even with 3,5" drives, but you would need to pull the entire enclosure out of the rack to physically remove or add a disk. –  the-wabbit Dec 13 '11 at 12:50
    
Thanks, I know that. But the example I made was tailored to our usual hardware, where we rarely need all those disks and would be a lot of money for a lot of underutilized hardware. Also, the question was more technology-oriented rather than product-oriented, i.e. it was a generic question about hdd tech. –  Luke404 Dec 13 '11 at 16:17

5 Answers 5

up vote 5 down vote accepted

The main differences would be heat/power/density, capacity, rotational speeds and seek times. You've outlined the power and density differences. Large form-factor disks have higher capacities available in a wider variety of rotational speeds. You can obtain 2TB and 3TB SAS midline disks in 3.5" where that's not currently possible on the 2.5" side.

However, the seek times are typically lower on 2.5" disks, versus comparable 3.5" drives. That's a potential performance advantage.

Also see: Performance differences between 2.5" vs. 3.5" SAS

share|improve this answer
    
+1 for apples-to-apples comparison (eg. 3.5" disk tech VS 2.5" disk tech as I was asking, and not number of disks per volume of space). I missed the lower seek times - which now I see should be obvious since the heads will have to do shorter seeks. –  Luke404 Dec 13 '11 at 16:25

A good discussion between 2.5" and 3.5" drives http://www.tomshardware.com/reviews/enterprise-hdd-2.5,2609.html

Quoted from the site:

"Translated into enterprise environments, it's easy to see that you can typically fit twice the amount of 2.5” drives into the rack space required by 3.5” drives. Blades don’t even support 3.5” drives because of their physical dimensions. Since capacities and I/O performance are almost identical between 3.5” and 2.5” products, but power consumption and dimensions are much decreased on 2.5”, the latter effectivly doubles power efficiency and storage density when performance per watt or capacity per space are concerned."

share|improve this answer

Smaller disks obviously mean a higher number of disks per rack - if you compare the same number of similarly-performing disks, you would not get significant performance improvement. And if you do not care about power consumption, heat / dissipation or mechanical vibration, you typically would not favor 2,5" disks over their 3,5" counterparts because of costs per gigabyte.

But looking at HDDs with their high seek times there is a simple rule: the more disks you have, the more independent operations your array can perform - since ideally, each disk is able to seek independently from the others. This translates directly to IOPS disk performance which is an important indicator for many common storage-intensive applications like virtualization or databases.

So the main reason for using 2,5" disks in arrays is actually the ability to pack more of them into an enclosure. This reasoning does not apply to SSD disks which do not suffer the high seek times of HDDs and can be designed to perform several operations simultaneously, so the number of disks would be not as much of an issue performance-wise.

share|improve this answer
    
+1 If price per GB isn't a concern, then this is a huge point in favor of 2.5" drives. –  Bigbio2002 Dec 15 '11 at 23:56

To add to the already states advantages of density, heat, power and seek times, the 2.5" disks are more reliable, due to lower RV (rotational vibration)

share|improve this answer

This is a difficult question.. It needs to be complimented by what you want to achieve.

From my own experience.. not quoting from any body else. but what i have done with my on time, equipment I can tell you this. I used new and/or fully functional equipment.

If a machine needs to boot once a few months- then yea 2.5" will be ok, put more ram in the machine and make it use more memory instead of grinding the HDD. Using a 2.5" for multiple boots does get slower once the disk get fuller and more fragmented, I used a 2.5" on a media centre- because it was a small disk and i could just pop in on top of my raid... but after 8 months I just put a 80GB 3.5" because the boot time was annoying me heavily. I did not believe that should have made a difference.. but it has!

If you going need to make mass storage , I am talking about like massive raids.. 4-8 drives per array, then the benefit of using 2.5" is that they use far less power during spin up, are much more silent, and take up allot less space. If you got them in raid 5, replacing a drive is easy peasy, and maximise the STORAGE amount.. but performance is less (not saying its rubbish.. but it does not always get to full potential. especially over time)

On major difference in 2.5" to 3.5" is DOLLAR per MB, 3.5" are available in 3TB drives.. the largest 2,5" is 1TB.. and the price difference is shocking. (as of Nov 2011)

@Luke404 - Thanks for pointing out to us that Enterprise 2.5" 15000RPM drives do exist. And yes they do, the reason I never came across one.. Price tag.

The other consideration is how many times you going to write to these drives?

Me, I set up a 4 drive array using RAID5.. and i dont care about MAX performance because i save massive chunks of data.. and that is it- I need to read maybe once year, but usually get deleted when it gets obsolete.

You want to use fast drives for operating systems that will be doing a bit of hard work, or SQL server, webserver, some sort of uPNP, etc.. then 2.5" just don't cut it here, you need to get 15000RPM drives like Raptors that are twice as fast in responding than a normal 3.5"

But that said.. the era of Solid State is flooding datacenteres and servers that really on reading the same data more than writing it. SSD's are amazingly fast at reading data- I was impressed at how fast my Windows 7 loaded after I installed.. but at the time (Early 2010) the small chunk write speed, made me scream and run around with scissors. Today the same SSD, at version 3 is extermly improved at writing and could be possibly ok to use a system drive. I how ever change my config and used this to install my games to, which was ideal, because you install a game once and read it many times.. I used Stalker for benchmarking, and of a Raptor 15K it was faster in loading levels but still like 30-70seconds, but from the SSD, it was a few seconds.. i could not believe it.. I could tweak up the graphics and it ran like a monster.

So there is an independent report- I tried not to be biased or to stick to standard.. so please dont down vote me if i made an error-- I just wanted to share my personal experience from years of tinkering..

share|improve this answer
    
please do note enterprise hard disks are available in 2.5" 15.000rpm –  Luke404 Dec 13 '11 at 16:21
    
I never knew that- hence base don my experience –  ppumkin Dec 14 '11 at 11:13

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.