Adaptec's Serial ATA RAID 2810SA
In the market for an inexpensive alternative to SCSI and parallel ATA technology? The solution may lie in SATA.
April 26, 2004
![Network Computing logo Network Computing logo](https://eu-images.contentstack.com/v3/assets/bltde8121fc52c5c8f3/blt3f3d0318f746b1c2/65a530e4187606040a1d8b8c/placeholder.png?width=1280&auto=webp&quality=95&format=jpg&disable=upscale)
We tested the controller using disks of various size and capacity to determine whether the controller was limited by the bus speed of the server, the speed of the disks attached or the RAID processor itself. We tested using 160-GB, 7,200-RPM Barracuda 7200.7 disks from Seagate; 7,200-RPM, 80-GB DiamondMax Plus 9 and 7,200-RPM, 250-GB MaXLine Plus II disks from Maxtor; and 10,000-RPM, 74-GB Raptor WD740 disks from Western Digital. We used DE400 removable drive enclosures from StorCase Technology to house the drives and the hot-swap capabilities of the controller.
This RAID Really Cooks
Using the open-source IOMeter benchmarking tool and a beta release of the commercial HD Tach benchmark tool from Simpli Software, we put the 2810SA controller through its paces under both Windows 2000 Server and Windows 2003 Server. No discernable performance difference was noted between operating systems. The 2810SA topped out at 134-MB-per-second read and 102-MB-per-second write performance when doing large 2-MB block transfers. Worst-case performance occurred when using 64-Kb block writes ranging from 6- to 16-MB per second.
Our tests revealed that while disk geometry and spindle speed had little impact on large block transfer tests, there was a significant performance delta on small block transfers. Our tests prove that the controller is not throughput bound by disk; rather, it's most likely bound by the read and write performance of the RAID hardware on the controller. Even so, the 2810SA benchmarks out as one of the fastest ATA controllers ever tested at Network Computing. In all test cases, CPU utilization hovered between zero and six percent.
One additional metric of note is performance during a volume rebuild: Using HD Tach, we were able to clearly demonstrate that the controller tops out around 60 MB per second when a rebuild process is in progress and that read performance on large block transfers suffers the most during a rebuild operation.
Icing on the Cake
After benchmarking controller performance, we turned to the management tools and features of the RAID controller. Adaptec's driver suite includes a Web tool called Adaptec Storage Manager, Browser Edition, which provides a Web GUI interface for controlling, monitoring and configuring the 2810SA controller.
We tested hot-swap capabilities by removing drives from the StorCase DE400 enclosure. The Adaptec controller detected the lost drive and automatically began rebuilding the array when a hot spare was available. The controller behaved exceptionally well, even though the DE400 enclosure does not support the SAF-TE enclosure management protocol. After replacing the drive, the controller can be instructed to manually rescan its SATA ports for new devices, thereby avoiding the need to reboot.
The 2810SA controller offers online expansion capabilities, but these capabilities are limited. The controller can grow an array by expanding it onto unused space on additional disks not already in the array. For example, if you have a four-disk array using 80-GB disks and want to expand that array, you could add four 250-GB disks and use 80 GB on each new disk to expand the existing array. In contrast, if you build an 80-GB array on four 250-GB disks, you cannot expand that array in place; the only way to expand an array is to add a disk to the system, even though space may be available on the existing disks in the system.
We also tested the ability to change RAID types on the fly. We discovered that the controller can rebuild any array into a different RAID-level with no data loss so long as the new RAID level uses equal or less disk capacity than the old array.
Adaptec Inc. Adaptec Serial ATA RAID 2810SA. Price: MSRP = $625. 800-442-7274 and 408-945-8600www.adaptec.com |
In addition, we tested Adaptec's optimized disk utilization feature, which prevents space waste when drives of different geometries are used in a single system. We built a RAID on four 80-GB Maxtor disks and then added four more 250-GB disks to the system, expanding the original RAID to include the new disks. In a traditional RAID system, we would have lost 170 GB of disk space on each disk we added to the volume. The 2810SA, however, allowed us to use the remaining space to build an additional array. This feature helps customers cope with the rapid evolution of SATA disk technology. But it comes at a price: If you choose to create multiple RAID volumes on a single disk using this feature, you increase your risk of data loss should multiple disks in the array fail.
Network Computing would like to gratefully acknowledge the contributions of the following vendors that provided the equipment and tools necessary to benchmark this product:
Simpli Software HD Tach Version 3.0http://www.simplisoftware.com[email protected]
StorCase Technologies DE400 Removable Drive Enclosurehttp://www.storcase.com
(800) 435-0642
Seagate Technologies ST3160021AS 160GB Barracuda 7200.7 SATA Disc Drivehttp://www.seagate.com(877) 271-3285
Maxtor Corporation DiamondMax Plus 9 80 GB SATA DiscMaxtor Corporation MaXLine Plus II 250 GB SATA Dischttp://www.maxtor.com(949) 347-7808
Western Digital Raptor Model WD740 74 GB SATA Dischttp://www.westerndigital.com(888) 935-8893
Joel Conover, a former senior technology editor of NETWORK COMPUTING is principal analyst for enterprise infrastructure at competitive intelligence firm Current Analysis. Write to him at [email protected].
You May Also Like
Maximizing cloud potential: Building and operating an effective Cloud Center of Excellence (CCoE)
September 10, 2024Radical Automation of ITSM
September 19, 2024Maximizing Manufacturing Efficiency with Real-Time Production Monitoring
September 25, 2024