|
|
Advertisement:
|
|
HighPoint Technologies e.SATA RAID Kit v2 |
|
Join the community - in the OCAU Forums!
|
Performance and Conclusions
PERFORMANCE
After we found the e.SATA RAID Kit working as advertised we finally wanted to see how the kit is performing in comparison with other SATA RAID solutions. For this purpose we compared it with Intel’s Southbridge integrated RAID solution and Silicon Images’ SiI 3114 4-channel RAID controller.
  
Intel’s ICH5-R was the first Southbridge implemented on desktop mainboards offering integrated RAID. Functionality however is limited to RAID 0 and RAID 1. RAID 0 (striping) does not offer fault tolerance and RAID 1 (mirroring) does not offer the performance advantage most desktop users have in mind when they set up a RAID system. The 2-channel Silicon Image SATA 3112 chipset is a very popular on-board solution implemented on many newer mainboards. Its bigger brother, the 4-channel SiI 3114 chipset we are using for this test, is currently only available on ABIT’s Max 3 motherboard. We have recently tested this mainboard here with good results and are using it again for this comparison. Below we show the RR1542 installed in the Max 3.
Before we present our test results we would like to offer some general thoughts about the RAID solutions in this test. A major difference between Southbridge integrated RAID as offered by Intel’s ICH5-R on one hand and the Highpoint RR1542 and the SiI 3114 on the other hand is that the latter two RAID solutions are using the PCI bus. The current 32-bit / 33 MHz PCI bus is limited by a relatively low maximum theoretical bandwidth of 132 MB/sec. RAID 0 with 2 SATA disks is in theory good for 300 MB/sec and such bandwidth is no problem for Intel’s ICH5-R used with the 876/875 chipsets. However, “maximum” interface bandwidth is in a true way only describing the “theoretical” maximum speed the interface would allow: SATA 150 x 2 = 300 MB/sec. The actual achievable transfer rate, however, is determined by the disk mechanics, the platter design, and the firmware. Even the currently fastest SATA disk, the 10,000 rpm Western Digital WD360GD “Raptor”, is not exceeding a sequential transfer rate of around 58 MB/sec on the outer (fastest) disk platter zone when running as single disk and around 110 MB/sec when running in RAID 0. Such transfer rates can still be handled by the PCI bus. But from the same point of view a PCI based 4-channel SATA 150 RAID controller does not make sense, because RAID 0 with 4 high-end disks like the WD360GD Raptors is by far exceeding the possible PCI bandwidth. But not only because of this do we think that Raid 0 with 4 disks is a less beneficial usage scenario for a PCI based 4-channel RAID controller like the RR1542. As we mentioned already: A Raid 0 array has no fault tolerance at all. Remember that a 4-disk array is even more likely to fault than a 2-disk array. Furthermore: Based on our earlier tests, Raid 0 with 4 disks offers very little real world performance advantage over a RAID 0 with 2 disks. We therefore recommend using the e.SATA RAID Kit either as external back-up/storage or for RAID 0/1, 5, or 1/0 arrays.
In our first test we compared Highpoint’s RR 1542 with Intel’s ICH5-R and the Silicon Image 3114 controller using a single Western Digital WD 360GD “Raptor” hard disk. We used the HDTach 2.61 hard disk benchmarking utility to take sequential read/write transfer rates and the disk access time. We then tested the Raptor with applications that are usual for a desktop environment. Unfortunately there is only the rather dated Winbench 99 test program currently available for application based HDD benchmarking. WB99 is running a collection of high-end desktop applications. They are “I/O bound” scripted in such a way that they are intensively reading and writing from and to the disk, by this they are “magnifying” performance differences between hard disk. WB99 comprises among others 2 benchmarking test sets, one representing a business environment and the other one a high-end desktop environment. In addition we ran Content Creation 2002 and Business Winstone 2002. They as well are simulating PC usage with business and content creating applications such a Lotus Notes, various MS Office applications, FrontPage, Photoshop, and various video and audio editing applications. The later 2 programs are measuring “system performance” and do not emphasise hard disk performance as WB 99 does. The test results show that all 3 competitors deliver almost identical performance. Differences between transfer rates, access time, and application performance basically do not exist.
In our second test run we used 2 Western Digital WD360GD in RAID 0. Again we can see that there is basically no performance difference between the Highpoint, Silicon Image and Intel RAID solutions. It is obvious that the PCI bus, although limited to only 132 MB/sec maximum bandwidth has no problem to handle a 2-disk RAID 0 array. As already speculated above the actual transfer rates with a 2-disk RAID 0 are far below the theoretical maximum of 2 x150 MB/sec with 2 SATA disks and obviously well within the capacity of the PCI bus. The RAID 0 performance with the 2 PCI based RAID adaptors is therefore not worse than with the ICH5-R which has a much higher theoretical maximum bandwidth (ICH to MCH) of 266 MB/sec.
  
Left to Right above: 2x WD360GD in RAID 0 on ICH5-R, SiL3114 and RocketRaid1542
Finally, we tested the RAID Kit’s external e.SATA connection using a single Maxtor Diamond Plus 9 PATA hard disk from the external disk enclosure. As reference all tests were repeated with the same disk running from Intel’s ICH5-R PATA channel. E.SATA is a relatively new and unknown Highpoint Technologies propriety standard. Our test results show that it works: We did not find any performance difference or penalty between the ICH5-R and the e.SATA connection running from the RR 1542.
SUMMARY AND CONCLUSION
Highpoint’s e.SATA RAID v2 kit is a combination of a 4-channel PCI SATA RAID card and an aluminium dual bay external harddisk enclosure. The kit allows running 4 hard disks in total: 2 SATA hard disks from inside the PC case and 2 PATA hard disks from the external dual bay disk enclosure. The disks in the external enclosure are connected to the SATA RAID adaptor via Highpoint's e.SATA propriety interface, while the internal SATA disk are using the RAID adaptors regular SATA 150 interface. It’s possible to run all of the 4 hard disks independently and boot into the OS from each of them or to combine them to RAID 0, 1, 0/1, 5, 1/0 and JBOD arrays. We found Highpoint’s e.SATA RAID Kit v2 well made, well performing, and highly versatile. It is providing increased application performance and data safety through several RAID levels that are possible only with a 4-channel RAID controller and is at the same time serving as an external back-up and storage solution. Furthermore it let the user conveniently run PATA disks from a SATA interface without resorting to expensive adaptors. These functions are far exceeding Southbridge integrated RAID, for example by Intel’s ICH5-R, or onboard PCI RAID solutions. The e.SATA RAID Kit v2 is relatively new and not yet available in all markets. The lowest price we found quoted is 210 US$ from NewEgg. That’s not exactly cheap, but still appropriate for the many functions the kit is offering.
|
|
Advertisement:
All original content copyright James Rolfe.
All rights reserved. No reproduction allowed without written permission.
Interested in advertising on OCAU? Contact us for info.
|
|
