Five RAID units tested
It's affordable and easy to manage -- two qualities you rarely hear mentioned about storage. We test your RAID options.
DAS is ideal for localised file sharing for a single or small number of servers -- the DAS is managed using the network operating system of the server it is attached to. While management and administration are initially quite simple, as the complexity increases with additional servers it becomes more difficult to manage as the storage for each server must be handled separately.
In terms of cost, DAS is very inexpensive when compared with network attached storage (NAS), however, if you are anticipating rapid growth bear in mind that DAS is limited in its scalability when compared to NAS.
Here is a list of things to look for when purchasing a DAS appliance:
- Scalability
- Fault tolerance
- RAID configurations supported
- Redundant power supplies and controllers
- Battery backup on controller cache
- Automatic failover
- Global hot spare drives
- Automatic and transparent drive rebuilding
- Management
- Remote management software
- Performance analysis and tuning tools
A few years back, if you pulled a drive from a RAID DAS unit it would have been a pretty fair bet that it would have been some flavour of SCSI. Recently, however, IDE has made strong inroads into DAS with the advent of powerful ATA RAID controllers. This has a two-fold advantage: not only are ATA drives less expensive than a similarly sized SCSI drive, ATA drives are also available in much larger capacities.
There is a downside of course. SCSI does have a higher bandwidth and is available with 15,000rpm spindle speeds compared to 7200rpm for ATA, it has shorter seek times and generally significantly longer mean time between failure (MTBF).
Another drive type that is also encroaching in the DAS space is Serial ATA (SATA), although SATA drives have similar advantages and disadvantages to the ATA.
Of the DAS units tested, the Adaptec has Fibre Channel SCSI drives, the ATABeast (as the name suggests) ATA drives, and the Promise comes with SATA drives.
Price:
AU$27,990 (14GB drive, 1.022TB, dual-controller configuration)
AU$17,349 (7-drive, 1.022TB, single-controller configuration).
Test Configuration:
14 x Seagate Cheetah ST336605FC Fibre Channel drives (0.514TB raw) in a RAID 5 configuration.
Drive specs:
- Capacity: 36.7GB
- Rotational speed: 10000 RPM
- Average seek: 5.1ms
- Buffer: 4MB
- MTBF: 1,200,000 hours
The test unit had a swag of redundant and hot-swappable components including two RAID controllers with battery-backed cache, dual power supplies, and two fan cooling modules, each with a small battery pack to ensure the enclosure is cooled even if all external power sources fail. This ensures your drives and components do not sit and stew in their own accumulated heat during a catastrophic power failure. The cooling module fans are surprisingly powerful and ensure that large volumes of air circulate around the drives.
Connection to the host is via two Fibre Channel loops providing 2Gb connectivity.
While each drive cradle has LEDs to indicate the drive's health and status including online, offline, fail, rebuild, drive identify, and prepare for removal, the SANbloc itself has status indicators for power, shelf faults, LS fault, FC loop open/closed, and FC loop speed.
The drive cradles and other redundant items such as power supply or cooling units are all quite easy to replace, and all items are hot swappable.
Management of the SANbloc is handled via Spheras Storage Director which uses Java RMI for connectivity over the Internet. Spheras is a very powerful utility that allows the user to configure, monitor, and maintain the SANbloc.
Adaptec was unable to provide the Lab with a 1TB implementation of the SANbloc and as a consequence the unit was tested with 14 drives (that amounted to just half a terabyte). For cost-per-GB calculations Adaptec supplied the Lab with pricing on a 14-drive 1.022TB solution rather than the cost as tested. Even so, the SANbloc is quite expensive at AU$27.39 per gigabyte -- but bear in mind that this price included the dual-controller option and 14 smallish drives. We also requested Adaptec supply us with a "best value" 1.022TB system with just a single controller and seven drives and the cost per GB drops dramatically to just AU$16.98 per GB.
Performance on the other hand was strong with the SANbloc the fastest during writes and quite quick at reads as well.
| |||||||||||||||||||||||||||||||
Price:
As tested AU$15,460
Test configuration:
7 x HP "branded" Model BD14685A26 Ultra320 SCSI drives (1.027TB raw) in a RAID-ADG configuration.
Drive specs:
- Capacity: 146GB
- Rotational speed: 10000 RPM
- Average seek: 4.9ms
- Buffer: 8MB
- MTBF: 1,400,000 hours
In terms of case volume, the HP is the smallest of all the units tested. Even though the case is 4RU the HP is very shallow in depth in comparison to the other units. Or at least it would be very shallow if it were not for the enormous pair of cooling fans on the dual power supplies that resemble rocket booster nozzles out of some cheesy sci-fi flick. The turbine fans are very large and any one of the fans is more than capable of providing adequate airflow through the power supply, controllers, and drive array. The test unit had good redundancy with two power supplies and two controller modules.
We should note that the unit shipped with a pair of HP Smart Array 642 Controller cards for the host server/s. We tried connecting the unit to our Dell's SCSI RAID controller but were unsuccessful as the 500 G2's driver software is written specifically for the HP Smart Array controller. It goes without saying, of course, that the host connectivity is Ultra320 SCSI and up to a maximum of four hosts are supported by the HP.
The array drives are Ultra320 SCSI units and while they are more expensive than the ATA alternatives they have a lower latency because of their higher rotational speed, and a higher MTBF in high I/O environments. The HP drive cradles are very substantial and constructed from cast aluminium with very smooth insertion and removal.
Management of the HP is the purview of the Array Configuration Utility (ACU) which has a relatively small subset of configuration and management tasks but was certainly more than adequate for basic maintenance, and there were some nice touches like the ability to finely tune the read/write cache ratio.
We should note, however, that some of the versions of this tool on HP's Web site will not work with the 500 G2. We initially downloaded V6.40 and were unable to "discover" the 500 G2 with the ACU. HP suggested we update to the latest version 7.16.2.0, which worked without a hitch.
Quite frankly, we expected the HP's cost per GB to be a little higher but it was a reasonable AU$15.05. We also expected performance to be lower than the other units as the HP's array was configured as RAID-ADG, which is in general not as fast as RAID-5. The HP's performance was below the Adaptec and Nexsan in the large file write test but the HP was only marginally slower than the fastest, the SANbloc, in the small file write test. In all the file read tests the HP was the fastest of all the units tested.
| |||||||||||||||||||||||||||||||
Price:
As tested AU$39,979
Test configuration:
28 x Maxtor Maxline II Model 5A300J0 ATA/133 drives (8.4TB raw) in a RAID 5 configuration.
Drive specs:
- Capacity: 300GB
- Rotational speed: 5400 RPM
- Average seek: 10ms
- Buffer: 2MB
- MTBF: 1,000,000 hours (but only in low I/O secondary storage applications)
We should also note that if you don't need such a high capacity, Nexsan also offer a smaller 3U "ATAboy2" with 3 to 14 drives and prices starting from AU$13,995.
One would expect 42 drives in a fully populated cabinet to generate quite a lot of heat, especially as there is not a lot of space between them, further restricting airflow. A large turbine fan in each of the three power supplies is the sum total of the cooling for the drives. The vendor assures us that this is more than adequate for the task, and we must admit the airflow through the cooling grills at the front of the case is surprisingly substantial.
The drives themselves, as the name of the unit alludes, are PATA, so they are relatively inexpensive when compared to SCSI for example. Connectivity to the host is via dual port 2Gb Fibre Channel controllers on each of the PATA controller modules.
The ATA RAID controller has a battery-backed cache and three power supplies. An optional second redundant ATA RAID controller can be fitted.
The front panel has a variety of Status LEDs for power supply and power supply blower status, controller and cache battery status, RAID and spares status, as well as a general fault indication.
Managing the ATABeast is very simple as the Web interface is excellent. The graphics and layout are one of the most colourful and user-friendly we have seen and nearly any function you would wish to monitor or reconfigure can be simply accessed from the interface.
Because the ATABeast featured 28 low-cost IDE drives that totalled a whopping 8.4TB it could be argued that this gives the unit an unfair advantage in terms of cost per gigabyte -- with 8.4TB on board the cost is only AU$4.76 per gigabyte. However, we should note that the minimum number of drives you can have in the ATABeast is 28. If, as we mentioned earlier, you do not need such high capacity the ATABaby2 would be a more cost-effective solution.
Overall performance was strong with the ATABeast posting the fastest write to disk with large files and although small file read and write operations were slower than the HP or Adaptec they were still quite impressive.
| |||||||||||||||||||||||||||||||
Price:
As tested AU$8822
Test configuration:
7 x Western Digital WD1600SD SATA/150 drives (1.12TB raw) in a RAID 5 configuration.
Drive specs:
- Capacity: 160GB
- Rotational speed: 7200 RPM
- Average seek: Read 8.9ms / Write 10.9ms
- Buffer: 8MB
- MTBF: 1,000,000 hours
Like EMC, Promise has chosen to populate the VTrack with SATA drives, which, while less expensive and available in higher capacities than SCSI, are nevertheless not as long lived in high I/O environments. This is of course less an issue in a redundant drive environment -- it just means you will have to swap out dead drives more frequently than with a SCSI-based system.
It is also possible to fit PATA drives to the system with an optional cradle adaptor.
The design employs quite a bit of redundancy with dual power supplies and dual cooling modules -- the latter are very impressive units each with a pair of large turbine-type fans that move a very large volume of air and there appears to be facilities for battery modules for each cooling module but these were not implemented in the test system.
There are a couple of single points of failure, one being the "Storage Enclosure Processor" and the other the SCSI controller. While the latter has dual Ultra160 SCSI channels they are both implemented on a single controller module.
The controllers have a 256MB cache, expandable up to 512MB, which is battery backed and provides up to 72 hours of protection in the event of a power failure.
Each drive cradle has a pair of status LEDs so it is simple to identify a faulty drive and the front panel of the unit has status and activity indicators for the SCSI channels, controller, and logical drive.
Administration of the VTrack is achieved with a product called WebPAM Professional in conjunction with your Web browser.
Configuring and managing the controller and drive array is very simple and straightforward using WebPAM.
In terms of cost per gig, the VTrack is a very cost effective solution at just AU$7.88 per GB.
System Performance, however was not the VTrack's strong suite -- it was the slowest in all the tests and is significantly slower than the other vendors in the large file transfer tests. Having said this, the performance of the VTrack is not actually "bad" and given the low total cost of the unit still represents good value.
| |||||||||||||||||||||||||||||||
Unfortunately the EMC Clariion arrived at the Lab well past the testing deadline. Still we did try to include the unit in our tests but the EMC's Fibre Channel controller and our Adaptec 9210LP Host controller refused to communicate. EMC was unable to source an alternative host controller before the completion of the comparison. We did, however, continue to evaluate the unit but have not obviously included any performance test results.
Price:
As tested AU$11,455
Test Configuration:
3 x Maxtor Maxline Plus II Model 7Y250M0 SATA/150 drives (0.75TB raw) in a RAID 5 configuration.
- Capacity: 250GB
- Rotational speed: 7200 RPM
- Average seek: 9ms
- Buffer: 8MB
- MTBF: 1,000,000 (but only in low I/O secondary storage applications)
Although to be fair, if a drive does die you simply replace it with another relatively inexpensive drive -- so cost wise this may actually be a cheaper alternative than a SCSI-based DAS unit.
Given the 2RU height the drives are obviously mounted horizontally. EMC was the only unit to configure their unit in this manner. The drive cradles are very simple in construction, although not as basic as the Nexsan cradles, and are quick and smooth in removal and replacement.
The drive cradles are covered by a removable, and lockable, front facia with power and fault status indicators.
The unit has a single controller but there is a dual-controller option available. Likewise, there is only a single power supply but a second redundant power supply is an option. While on redundancy, the connectivity to the server is via a pair of Fibre Channel ports but there is an option for an additional two Fibre Channel ports. In the configuration sent to the Lab there is very little in the way of failover other than the RAID drives and battery-backed cache but should you require, further redundancy can be optioned.
Cooling is excellent with no less than five fans providing airflow to the controller and drive bays, each fan with a baffle to prevent dust entry when the unit is powered down, while the power supply also has a single fan. Should a fan fail it is a simple user-replaceable item.
A Web-based management tool called Navisphere is used to manage the Clariion and while the interface is perhaps not at "pretty" as some of the other vendors it nevertheless has a good deal of functionality.
Cost per GB is reasonable, although given the use of inexpensive SATA drives we would have expected the EMC to be less expensive than the SCSI drive configured HP for example but the EMC was slightly more expensive at AU$15.27.
| Product | Adaptec SANbloc | EMC CLARiiON AX100 | HP StorageWorks Modular Smart Array 500 G2 |
| Company | Adaptec Australia | EMC | HP Australia |
| Phone | 02 8875 7874 | 02 9922 7888 | 13 13 47 |
| Web | www.adaptec.com.au | www.emc.com | www.hp.com.au |
| RRP as tested (inc GST) | AU$27,990 as 1TB Dual controller unit (using 14 x 73Gb 10K drives) | AU$11,455 | AU$15,460.00 |
| Capacity as tested (GB) | 1022 | 750 | 1027 |
| Price Range (inc GST) | AU$17490 To AU$33190 | AU$9000-AU$19,500 | AU$8600 to AU$15,100 |
| Warranty & support (includes telephone, e-mail, Web details, and hours of support) | 3 years return to base (5 years on disks). 3 year on-site available | 24x7 phone support (1800 number) + Web support, customer / partner installable and maintainable | 3-year limited warranty next day on-site, including hardware diagnostic support, pre-failure warranty (coverage of hard drives, memory and processors). Enhancements to warranty services are available through HP Care Pack services. |
| RAID supported | 0, 1, 0/1, 3, 5, 10, 30, 50 | RAID-5 | 0, 1, 0+1, 5, ADG |
| Number of drive bays | 14 bays | 12 bays with 160GB or 250GB SATA drives | 14 hot swap |
| Controller type and number | Fibre channel (Dual redundant) | Single SATA controller upgradeable to Dual SATA | Dual Ultra320 SCSI controllers |
| Controller cache battery backed? | Yes | Yes (1GB battery backed cache) | Optional battery-backed cache of 512MB (standard with 256MB cache without battery backup) |
| Drive type | Fibre channel SCSI | 160GB or 250GB SATA Intermix | Ultra320 SCSI |
| Storage capacity min/max | Min 250Gb ( 14 x 18GB drives), max 16.4TB (14 x 8 shelves x 146GB drives) | 480GB (3x160GB) / 3TB (12x250GB) | Min (1x36GB HDD) 36GB to Max (14x300GB HDD) 4.2TB |
| Type/s of server connectivity supported | Dual-port 2Gb fibre channel controller | Up to 16 hosts (single connect) or 8 hosts (dual connect) | Dual Ultra320 SCSI channels for two hosts upgradable to support up to four hosts |
| Redundant power supplies? | Dual redundant hot swap power supplies | Dual power with dual controller option | Yes, dual |
| Size | 3RU | 2RU | 4RU |
| Other features | Redundant hot-swappable components with no single point of failure system. | Capability to take snapshots of volumes for backups without outages etc. Logical path management and failover. | Upgradable to fibre channel connectivity, redundant data controllers |
| Product | Nexsan ATABeast | Promise Technology Vtrak 15100 |
| Company | Busi-soft Computers - Nexsan Technologies | Westan |
| Phone | 1300 888 602 +1 818 715-9111 | 1300 78 03 03 |
| Web | www.busi-soft.com.au www.nexsan.com | www.westan.com.au |
| RRP as tested (inc GST) | AU$39,979.00 | AU$8822.00 |
| Capacity as tested (GB) | 8400 | 1120 |
| Price Range (inc GST) | AU$37,979 To AU$74,979 | From AU$7590 |
| Warranty & support (includes telephone, e-mail, Web details, and hours of support) | 3-year advanced replacement (on-site and other warranty upgrades available) | 3 years RTB included as standard. Business hours or 24x7 on-site warranty support available. |
| RAID supported | 0, 1+0, 4, 5 | 0, 1, 3, 5, 10, 50 |
| Number of drive bays | 42 hot swap bays | 15 hot swap |
| Controller type and number | Proprietary IDE (with second auto fail over controller optional) | SATA |
| Controller cache battery backed? | Yes | Yes (256MB expandable to 512MB with battery backup for up to 72 hours) |
| Drive type | ATA | Serial ATA (optional SATA to PATA adaptors available) |
| Storage capacity min/max | 7TB / 16.8TB (with 400GB drives) | Up to 5TB |
| Type/s of server connectivity supported | Dual-port 2Gb fibre channel controller (dual controllers with auto failover optional) | Dual ultra 160 SCSI (can support up to two servers) |
| Redundant power supplies? | Three redundant hot swap power supplies | Dual hot swap |
| Size | 4RU | 3RU |
| Other features | 10/100TX port for remote management. Scales up to 420 drives per rack (10 ATABeasts per rack). Configurable for no single point of failure. | Redundant cooling system, up two 8 VTraks can be daisy chained together on a single SCSI channel, Remote management software. E-mail notification, SNMP. |
Our test rig consisted of a host server and the RAID DAS unit under test. The host configuration was:
- Dell PowerEdge 6600
- Quad Xeon processors
- 3GB of DDR
- Adaptec Fibre Card 9210LP
- HP Smart Array 64X Controller
- Windows Server 2003
Large File Test
A 1,812,838,470 bytes (1.68GB) ZIP file that cannot be further compressed
Small File Test
A folder containing 4,567 files in 227 folders with a total size of 844,952,770 bytes (805MB). In each case the time taken to transfer the file to and from the Dell server from the RAID DAS unit under test was recorded and the throughput calculated.
Fibre Channel (FC) is a high performance interface designed to boost performance and the flexibility of storage systems with multiple drives.
Fibre Channel has several key benefits:
- Hot Pluggable -- drives can be installed or removed while the host is running.
- Performance -- the fastest interface in multi drive environments.
- High data integrity
- Longer cable lengths -- relative to LVD
- Cost effective -- other high end solutions require expensive adaptors
- Longer cables -- FC is 30m over copper and 10km over optic fibre while the total length for LVD is 12m
- Data redundancy -- frames include a CRC while SCSI simply relies on parity
- Dual Loop capability -- Allows two servers to be connected, one on each loop, at the same time so servers can share a drive without any manual switching
So when would you use FC in preference to LVD?
With five drives or less, FC offers no performance benefit over LVD. However, as the number of drives grow above five, so does the performance advantage of FC. FC currently peaks at 100Mbytes/sec while LVD stops at 80Mbytes/sec. A FC loop can also cope with up to 126 devices compared to LVD's 15 device limit. A dual FC loop, therefore, can accommodate up to 252 drives but there is a caveat. If two servers are connected to the DAS -- one per loop, there can be 252 devices provided the servers do not share any drives. Each shared drive accounts for a single device in both loops. So if all the drives are shared between two servers a dual loop can only support a maximum of 126 drives.
RAID is an acronym for Redundant Array of Inexpensive Discs and was originally conceived utilising SCSI Drive controllers. Initially SCSI interfaces were the only controllers that had the ability to control more than two drives but now multi-channel IDE and Serial ATA (SATA) controllers are quite a common and, at times, an inexpensive alternative to SCSI.
There are quite a few types of RAID arrays, one of which is a non-redundant array that offers greater performance but no data protection.
RAID-0. This array type utilises striping across two drives for example to provide better performance than two individual drives. There is no redundancy of data and therefore no fault tolerance.
RAID-1. Commonly known as disk mirroring, RAID-1 consists of at least two drives each pair of which contains an exact duplicate of each other. There is no striping but should one of the drives fail the second drive in the pair holds a full copy of the data. Read performance is actually improved as either of the drive pairs can be read at the same time although write performance is the same as a single drive as pairs must be written simultaneously.
RAID-2. This is rarely used now as the ECC information stored on the disks in the array has to some extent been supplanted by modern hard drives that include their own hardware ECC.
RAID-3. In this situation data is stripped and one of the drives in the array is dedicated to storing parity information. I/O cannot be overlapped as any I/O operation addresses all drives at the same time as a result RAID-3 is relatively slow.
RAID-4. The stripes used in RAID-4 implementations are relatively large so records can be read from a single drive and drive read operations can be overlapped to improve speed. However there is no overlapping of write operations as the parity drive must be updated with every write operation.
RAID-5. An improvement on RAID-4 as it includes a rotating parity array which means that all read and write I/O can be overlapped. RAID-5 does not store redundant data but by using the parity array information data can be reconstructed when a faulty drive is replaced. RAID-5 requires a minimum of three drives but typically five drives in an array.
RAID-6. Is actually Dual Parity and takes the parity region from RAID-5 and duplicates it so each disk has two parity regions that are calculated separately. As a consequence a RAID-6 array can recover from the loss of two drives, but performance is impacted and an extra drive is required which increases the cost of implementation.
RAID-ADG. Or Advanced Data Guard is HP's proprietary implementation and is similar to RAID-6 in that RAID-ADG has Dual stipes on each drive.
RAID-7. Technically not an official RAID level RAID-7 is a proprietary system that includes an internal CPU, real-time embedded operating system and caching in the array hardware. The approach improves upon RAID-4 by managing the array asynchronously and thus improving random read and write performance over RAID-4. A large cache is critical to the systems performance and must be supported by UPS to prevent data loss should a power failure occur.
RAID-10. This type offers a combination of RAID-0 (striped) and RAID-1 (mirroring). As a consequence the array has high fault tolerance of mirroring and the speed of a RAID-0 striped array. But the implementation can be expensive as twice as many disks are required as a standard RAID-5 implementation also scalability is limited.
RAID-53. This type combines striping by byte and parity across two disk sets and is fault tolerant and fast, particularly with large files. It is expensive to implement and like RAID-3 has poor utilisation of disk capacity.
We often hear from people who are surprised that it is possible to cool 42 drives in a compact 4U enclosure, and it is a reasonable concern.
However, I can assure you that my design team spent many months experimenting with different configurations until we arrived at a design which would work.
Please allow me to point out a few things:
- ATA drives run much cooler than legacy Fibre and SCSI drives.
- There are three powerful blowers in the chassis and in fact we have tested the cooling with one blower failed at maximum workload and at the maximum allowable ambient temperature of 35C, to ensure that the disk HDA's never exceed the maximum recommended temperature in this failed state.
- Under normal operations, you can take off the lid and feel the drives to see for yourself that they are running at a reasonable temperature.
- Note that the airflow is unrestricted in the drive cage, as the drives are mounted vertically. If you have access to a smoke generator, you can see that the air flows between drives at a remarkable speed (thermal transfer is proportional to air velocity)
- Note that the spacing between rows of drives is asymmetric -- a design tactic which optimises thermal transfer from the drives to the airstream.
- We use blowers instead of fans, as when you have a design with high static pressure, only blowers can maintain their rated CFM (fans, commonly used in cheaper enclosures, lose efficiency if airflow is impeded in any way).
External-storage market continues to grow
Factory revenues for external disk storage systems grew 3.5 percent year-over-year to AU$4.4 billion in the third quarter of 2004, according to data released by IDC this week.
This is the sixth consecutive quarter of year-over-year revenue growth, according to the research company. The total disk storage systems market grew year-over-year as well, climbing 2.1 percent. Storage capacity also increased, by 50.5 percent year-over-year to 310 petabytes -- the largest growth rate over the past seven quarters.
EMC claimed the top spot with 21.2 percent of revenue share in the external disk storage systems market, posting a 17.4 percent year-over-year gain in revenue during the quarter.
Second place HP, which had 19 percent of the market, saw revenue dip 7.5 percent year-over-year. IBM came in third with 13.1 percent market share. NetApp posted the strongest year-over-year revenue growth during the third quarter, with 24.6 percent growth, says IDC.
EMC was also on top in the total networked storage market with 28.9 percent revenue share, followed by HP, with 23.3 percent, and IBM, with 11.5 percent. For the top ranking in the Open/iSCSI SAN market, HP and EMC differed by only 0.2 percent, the market researcher says.
IDC attributed the growth in the external disk storage market to increased adoption of entry and midrange networked storage, which it defines as storage priced below AU$150,000 per system. The top vendors introduced external storage systems and solutions targeted to specific customer and application segments during the third quarter.
In the network attached storage, or NAS, market, which grew 14.3 percent over the third quarter of 2003, NetApp led with 36.3 percent revenue share, followed by EMC with 34.8 percent share. HP was in the top position in the total global disk storage systems market with 23.6 percent revenue share, while IBM was in the second place with 20.6 percent share.
"We are encouraged to see continued growth in disk storage systems, despite the lower rate when compared to previous quarters," Brad Nisbet, program manager at IDC, says. "Conversely, we saw an increase in the growth of petabytes shipped, which is yielding the largest dollar per gigabyte pricing decline in seven quarters and points to a growing share of higher-capacity, lower-cost disk drive deployments and a broader variety of products offered by the major vendors."
Nexsan ATA Beast
If we stay precisely true to the scenario then it is a close tussle between the Adaptec SANbloc, with single controller and 7 drives, and the HP Modular Smart Array 500 G2 with the HP just winning on price. Unless, of course, rack space is at a premium and then the 3RU SANbloc appears to become the more attractive option over the 4RU HP, except the HP has the option of higher capacity drives and thus trumps the SANbloc in terms of raw capacity.
However, if our 1TB scenario is just your starting point and you expect quite rapid growth in your storage requirements then the initial purchase price of the Nexsan ATABeast becomes less of a concern as the ATABeast is, by far, the least expensive of the units in terms of price per gig with its ability to swallow up to 42 inexpensive PATA drives in a 4RU form factor.
Company: Finn Productions
Scenario: This company's database server is running out of storage space. The company wants to add on additional storage to the server.
Approximate budget: Open
Requires: A direct-attached storage device with at least 1TB of storage.
Concerns: Performance of the storage device will be important, as well as the ability to add more capacity in future if required.