<!–#set var="article_header" value="IDE-RAID With Notebook Drives:
Quick And Quiet” –>
RAID With 2.5″ Hard Drives: Quick And Quiet
The article first appeared in Tom’s Hard News Email Newsletter
For many of us, there is a constant desire to increase the performance of our PC. To this end, a high-speed processor, a generous amount of RAM and a reliable motherboard are necessities. However, the hard disk drive plays a very important role, as well. For instance, whether you want to run a huge application or start Windows, both the system and the user have to wait for the data output from the disk drive – this slows us down.
Over the years, the industry has brought us higher capacity and faster drives. Yet they come at a price: the disturbing noise generated while idling and accessing have been the center of much discussion, leading to significant efforts towards reducing the annoyance. This is especially important because the powerful fans needed for todays top PC systems (used to cool the CPU, the graphics chip and the power supply unit) only contribute to the acoustic harassment.
Choosing your components wisely is still the best way to keep the noise level to within an acceptable range. But one of the primary sources of noise, the hard drive, will be with us for many years.
Nevertheless, there are quieter models. Notebook drives (2.5″ models) run at reduced rotation speed and are therefore very quiet. The only disadvantage involved is that they have much lower performance.
Usually, the disk subsystem performance is increased by setting up RAID arrays with several drives. Data is equally spread to all drives (so-called striping), resulting in much higher performance due to the simultaneous activity of all drives. Such a RAID 0 array can of course be established with 2.5″ – in the end, they are nothing more than mini IDE drives. In this article, we take a closer look at the performance level that can be reached by such an array.
IDE-RAID With Mini Hard Drives: THG Shows You How
So far, we’ve not heard of anybody putting 2.5″ drives into a RAID array. However, all we need is a simple test system, an IDE RAID controller and two notebook drives. Let’s get started!
We received two 2.5″ drives from Fujitsu: the MHN2150AT and its bigger brother, MHN2300AT. The two drives are equals, technically speaking, only differing in capacity (15 vs. 30 GB) and some performance differences.
Fujitsu MHN2150AT And MHN2300AT: 2.5″ At Its Best
Here’s a summary of features:
- 4.200 rpm
- 2 MB Cache
- 12 ms Seek Time
- UltraATA/100 Interface
- Size: 70 mm x 100 mm x 9,5 mm
The drives look pretty much the same, and both will tell you the configuration information. Here, what applies to desktop drives can be applied to notebook models: they can be run either as Master (drive 0) or as Slave (drive 1).
Refer to Fujitsu’s Website for more information about these drives:
http://hdd.fujitsu.com/global/drive/mhn2xxx/catalog.html
This picture makes it quite clear how small such a notebook drive is, in comparison to a 3.5″ diskette.
The 2.5″ RAID-Array
As already mentioned, we still need an IDE RAID controller. Models from HighPoint (HPT372) and Promise (PDC20265R) are widely used. Both of them can be found either as on-board components on an increasing number of motherboards, or as stand-alone PCI cards (see picture above). Our IDE RAID Controller Test did not reveal significant differences for general purpose use, so your decision can be based on the prices. You can find more information about both controllers on the following websites:
http://www.promise.com
http://www.highpoint-tech.com
Installing 2.5″ Drives Into Desktop Computers
2.5″ drives are made for notebooks, so how can you install them properly into your home computer? There are two things that are required: first, we will need an adapter, since the connector of notebook drives unifies both data lines and the power supply; second, we need to get mounting frames for the hard drive. There are such frames for 3.5″ that are to be installed into 5.25″ bays – so there must be frames for converting 3.5″ to 2.5″.
This image shows an adapter which splits the data lines from the power connector. It is the only piece of hardware that is required to run 2.5″ drives with any IDE controller. At the right side you can see the mounting frames for installing drives into a desktop computer.
After installing the frames, you can finally install the drive into a 3.5″ bay.
Test Setup
Hardware | |
CPU | Intel Pentium 4, 2000 MHz |
Memory | 256 MB DDR-SDRAM, PC2100, 133 MHz, CL2.0, Infineon |
Motherboard | ABit BD7-RAID (i845D, HighPoint HPT372 RAID) Intel 82801BA w/ UltraATA/100 BusMaster IDE Controller |
Graphics Card | GeForce 3, 64 MB DDR-SDRAM |
Drivers & Software | |
IDE/AGP Driver | Intel Application Accelerator 1.12, Intel BusMaster IDE Driver 6.20 HighPoint Driver Version 2.3 (25.12.2001) |
Graphics Driver | nVIDIA Detonator 4 Series, Version 23.11 |
DirectX Version | 8.1 |
Operating System | Windows 2000 SP2 + SRP1 |
Benchmarks | |
Data Transfer | HD Tach 2.61 from TCD Labs |
Application Performance | Ziff Davis WinBench 99 Version 2.0 Business and HighEnd Disk WinBench 99 |
Access Time | ZD WinBench 99 2.0 – Disk Inspection Test |
CPU Load | ZD WinBench 99 2.0 – Disk Inspection Test |
To get an idea of the performance level of the 2.5″ drive array, we also included an actual desktop drive. IBM’s DeskStar 60 GXP (IC35L040AV) stores 20 GB per platter, runs at 7,200 rpm and comes with 2 MB buffer. It is considered an excellent mid-range drive today.
Data Transfer Performance Diagrams
Fujitsu MHN2150AT
Fujitsu MHN2300AT
As expected, both 2.5″ drives provide just about the same performance. The maximum sequential read rate is about 20 MB/s and stays at > 10 MB/s at the end of the medium – that’s quite a good result for a notebook drive.
2x Fujitsu MHN2150AT – RAID Mode 0
HD Tach 2.61 doesn’t seem to like our 2.5″ drive array, because the results varied quite a bit. That’s why we had to switch to the Disc Inspection Test of WinBench 99 2.0.
Read Burst at 102.8 MB/s
The Read Burst Performance is nothing more than the maximum transfer rate (in MB/s) that can be delivered by the drive or the RAID array. This data is usually read directly from the drive’s buffer memory. IBM’s DeskStar GXP seems to use faster memory chips, as the performance is faster by about 10 MB/s. The RAID array does double this value, but performance increases only by 60%.
Access Time: Clearly Lags Behind Desktop Drives
The access time of the RAID 0 array is a bit quicker than that of a single notebook drive. Still, it is a far cry from the excellent results of the DeskStar 60GXP.
It should be mentioned that these results do not reflect the seek time that is usually advertised by manufacturers and computer shops, but the real access time. It includes head positioning and rotational latency, and this takes considerably longer than the theoretical time it needs to locate data.
Business & HighEnd Disk WinBench 99
Basically, both the Business and the HighEnd Disk WinBench 99 belong to the system benchmark WinStone 99. Here, the benchmark evaluates how long the system/ drive is occupied with the delegated tasks. In the end, you get a value that is useful for drive comparisons.
While a single MHN2150AT or MHN2300AT cannot possibly keep pace with the fast IBM DeskStar 60GXP, the application performance clearly increases in both benchmark runs. Still, the access time has quite a big impact on the result, as the pure data transfer rate is at the same level as the desktop drive. With frequent random access, notebook drives are clearly slower.
CPU Load: Exemplary
The IBM desktop drive evidently plays in a completely different league, but these performance numbers are brought about by a CPU load that is six times higher than the load of the 2.5″ drives: 6.7% vs. less than 1%. Of course, this does not really have an impact with today’s processing power, but it shows very clearly that a fast IDE drive really works the CPU!
Surface Temperature: A Cool Affair
This is one of the great characteristics of notebook drives – they do not become hot at all. Thanks to the slow rotation speed of 4,200 rpm, the surface temperature of the 2.5″ drives clearly remains below 40°C. Also, 2.5″ drives only require a fractional amount of the power that is consumed by desktop drives. Spinning up the medium sometimes requires more than 35 Watts in standard hard drives, while notebook models can live with 15 Watts.
Conclusion: Great Performance With 2.5″ Drives
The benchmark results are quite clear: only two 2.5″ drives are able to outperform a modern desktop drive in terms of transfer performance – and without the high temperatures and obnoxious noise! Using three or more drives will take up the full bandwidth of any IDE RAID controller without increasing the noise level. This is an ideal configuration for users that want a computer which is both fast and quiet. Take, for example, the multitude of home servers that are connected to the Internet via DSL or ADSL – typically, these computers are situated in bedrooms or living rooms. Here, a quiet PC is invaluable.
This could be a way for notebook manufacturers to add value to their top models. What about a notebook that runs two 2.5″ drives in an array (likely RAID 1) by default? Or imagine a docking station that comes with a second hard drive that automatically updates the data, resulting in excellent data security and significantly enhanced read performance. Ideas abound in this area.
As is often the case in life, the best things have their price – and this also applies to our 2.5″ RAID array. In addition to the cost of the RAID controller, you will have to obtain two notebook drives that start at approx. $120 for a 20 GB model. Two drives will cost you at least $240, which is roughly the price for a single 80GB or 100GB desktop drive. And finally, consider the fact that a quiet drive is only useful if other components in your system are not overwhelmingly loud.