|Introduction & Specifications|
The original Drobo direct-attached-storage (DAS) device hit the scene in summer of 2007 and took the tech world by storm. It represented a new paradigm in how you could implement external, redundant back-up storage with much of the same goodness that RAID offers, but without a number of the headaches and the complexity that also comes with traditional RAID solutions. The Drobo still used RAID to redundantly store data across multiple hard drives, but did so using Data Robotics’ own version of it, which it calls BeyondRAID (see the table below). The primary advantages that Data Robotics’ BeyondRAID has over traditional RAID is that with BeyondRAID you can add and remove hard drives that have different capacities and speeds and from different manufacturers; and you do this while the device is operating--without any downtime--and without needing to do any manual data migration.
Drobo S, which is a five-bay DAS device, with eSATA, FireWire 800, and USB 2.0 ports. (The latest iteration of the Drobo--not the newer "S" model--has four bays and only FireWire 800 and USB 2.0 ports.) The Drobo S has another unique advantage over the Drobo--and even all but the most robust RAID solutions--in that, with at least three drives installed, the device’s data can be protected from the failure of up to two drives.
The Drobo S is targeted primarily towards "creative pros, photographers, videographers, small office/home office" users, and is available from a healthy number of suppliers in both bare-bones and populated versions. A unit with no drives has an MSRP of $799--which is certainly on the high end when you consider that you can purchase a simple, 2TB external hard drive for less than $150 these days. But a plain old external hard drive won’t give you anywhere near the same amount of data integrity and expandability that the Drobo S offers.
|Setup and Features|
The Drobo S can work with versions of Windows going all the way back to Windows XP; it can also work with Mac OS X 10.5 and newer; and the Drobo S even has beta support for Linux. The first step in setting up the Drobo S is to install the Drobo Dashboard software on your system--there are Windows and Mac versions of it. Once the software is installed and running, you next install at least two SATA hard drives into the Drobo S device--the drives slide right into slots on the front of the device, without need for any rails or tools. You then power up the unit and attach it your system via the eSATA, FireWire 800, or USB 2.0 connectors on the back of the device.
Depending on your system, you might run into some compatibility issues with the eSATA or FireWire connections. For instance, initially whenever we attached the Drobo S to our Windows 7 testbed via the eSATA connection, our system always crashed, displaying a BSOD (this was the first time we ever experienced this issue with an eSATA device attached to our testbed). So we swapped out the eSATA controller card with one that Data Robotics reported was known to be compatible with the Drobo S, and then all worked fine. We ran into a similar issue with FireWire where we couldn't get our system to see the Drobo S using our testbed's built-in FireWire connection, and instead had to use a FireWire add-in card. If you want to use the Drobo S with a system’s FireWire 400 connection, you’ll need to purchase a FireWire-400-to-FireWire-800 adapter (which you can typically find for around $10). If you use the eSATA or FireWire connections you also might need to experiment with the order in which you connect the Drobo S to your system--in some cases, the Drobo S needs to be powered on and connected before your boot up your system; and in other cases, you might need to wait to connect the Drobo S until after your system has booted into the OS. That said, the USB 2.0 connection appears to work flawlessly, without any issues at all.
Once the Drobo S is successfully connected to your system and the Drobo Dashboard senses that the device is attached, you’ll see a message telling you that the device’s drives aren't formatted, and asks you if you want to format them. On a Windows system, if you click yes, you are taken to a screen that gives you the option of formatting the drives using a Windows XP-compatible version of NTFS (which limits volume size to 2TB), a Windows Vista (and newer)-compatible version of NTFS (supporting up to a 16TB volume size), or FAT32 (also limited to a 2TB volume size). (On the Mac, your choices are HFS+ and FAT32 file systems. If you want to format the drives as EXT3 for a Linux system, you need to follow the instructions here). Once you pick your file format, you then choose the size of your volume, name the volume, and map a drive letter to it. After you've made these selections, the Drobo S goes to work formatting the drives to appear as a single volume. You can also set up multiple volumes if you want. If you see a Windows OS message pop up saying "You need to format the disk in drive x: before you can use it. Do you want to format it?", ignore it and click cancel.
Once formatting is complete--which should only take a few minutes--the Drobo S is ready for action. By default, the Drobo S is protected against the failure of a single drive. You can, however, choose the "Dual Disk Redundancy" option to protect against the failure of up to two drives. If you choose Dual Disk Redundancy, however, you will have less available storage capacity. With our unit, we installed five lightning-fast 2TB Seagate Barracuda XT hard drives--when configured for single-drive protection, we had 7.22TB available for data storage, and 1.86TB reserved for protection; when configured for dual-drive protection, we had 5.4TB available for data, and 3.69TB for protection. You can switch back from dual-disk to single-disk protection if you want, but this conversion can take quite a bit of time (i.e., hours), depending on how much data is already stored on the drives--but if you need to access it during this conversion you can still read and write to it. Even when a drive fails, you still have full access to the device and the data stored on it--or even two failed drives if you are using Dual Disk Redundancy.
The front of the device has a set of LEDs that displays color-coded health information for each installed drive. The lights will be green, orange, or red, and solid or blinking, depending on the health of individual drives or the array--such as warning of a failed drive, a drive that is almost full, or that the device is in process of rebuilding the data protection. The Drobo Dashboard will also alert you if there is a problem. You can even configure the Drobo Dashboard to send you e-mail alerts if anything out the ordinary happens with the device. The Drobo Dashboard also includes a utility for setting up multiple, automated data backups.
With our Drobo S configured for Dual Disk Redundancy, we removed two of the five installed hard drives. Lights started flashing and warning screens popped up, but the device kept working--we could still read and write to it and none of our data was lost. We plugged in replacement drives and the Drobo S automatically started to rebuild the array and resetting data protection. In our experiment, it took the Drobo S about six hours to rebuild the array; and during this time, data transfer performance was noticeably slower.
At any point you can add more drives to the device if all five bays aren’t already populated--and the Drobo S will automatically format the new drives and add them to the array. You can also remove a drive and replace it with a larger capacity drive and the Drobo S will once again perform its automated magic. Try this with a traditional RAID device and you won't be very happy with the results.
|Test System and ATTO Disk Benchmark Testing|
Our Test Methodologies: The Drobo S and comparison devices were tested as external volumes attached to our HP Pavilion m9550f testbed system (2.5GHz Intel Core 2 Quad Q9300; 8GB PC2-6400 DDR2 SDRAM; 1TB NTFS 7,200RPM SATA 3Gb/Sec hard drive; ATI Radeon HD 4850 512MB, Windows 7 Ultimate 64-bit). We tested each device using its fastest (eSATA) and its slowest (USB 2.0) connection interfaces. All the devices were tested using 2TB Seagate Barracuda XT hard drives, formatted as a single NTFS volume, using a GUID Partition Table (GPT). We also tested each device in two modes: using its most aggressive and its least aggressive tolerance settings.
As our m9550f testbed does not have native eSATA support, we used an IOGear GICe711S3 PCI Express card. We didn't conduct formal performance testing with any of the devices' FireWire connections, but we did do anecdotal compatibility testing with those devices that had FireWire 400 and 800 connections, using a Vantec FireWire 400/800 PCI Host Card (UGT-FW100) in our testbed.
Networking, Windows firewall, automatic updates, and screen savers were all disabled before testing. With all test runs, we rebooted the system and waited several minutes for drive activity to settle before invoking a test. All tests were run multiple times (no less than three times, and often more) and we used the best (non-anomalous) score available.
We compared the Drobo S against the two other multi-drive DAS devices: the Vantec NexStar MX (NST-400MX-SR) and the Promise SmartStor DS4600. The NexStar MX is an inexpensive (around $100) two-bay device; with eSATA and USB 2.0 ports; and support for RAID 0, RAID 1, and JBOD. Even though the NexStar MX isn't in the same class as the Drobo S, we included it because we wanted to show the relative performance between the Drobo S and a much less-expensive alternative. If you don't need the storage capacity or feature set of the Drobo S, there are much less-expensive options available, such as the NexStar MX. The DS4600 represents a more apples-to-apples comparison with the Drobo S: It is a four-bay DAS device; with eSATA, USB 2.0, FireWire 400, and FireWire 800 ports; and support for RAID 0, RAID 1, RAID 5, and RAID 10. The DS4600 sells for around $300.
ATTO is a disk benchmark that measures transfers across a specific volume length, measuring raw transfer rates for both reads and writes. The charts below reflects a best-case performance scenario, using 8,192KB (8MB) transfer sizes over a total volume length of 256MB. Smaller transfer sizes in this test exhibit lower peak performance than what is reflected in the graphs below.
The most glaringly obvious and fully expected observation here is that the eSATA transfer rates of all the devices far exceed that of their USB 2.0 speeds. The maximum theoretical data transfer rate for USB 2.0 is about 60MB/Sec, which is notably slower than the actual transfer rate capabilities of many hard drives--meaning that the USB 2.0 interface itself will be the bottleneck in most instances. The eSATA interface we used has a theoretical data transfer rate of about 375MB/Sec, which is faster than any data transfer rate we've ever measured for a SATA hard drive--or even standard SATA SSDs currently, for that matter. For instance, the 2TB Seagate Barracuda XT drives we used in all the devices here, scored a write speed of 143.8MB/Sec and a read speed of 147.5MB/Sec on the ATTO test in our recent 2TB hard drive roundup. If your system has an eSATA interface, you'll see marked higher transfer speeds over USB 2.0. Both the Drobo S and the SmarStor DS4600 also support FireWire 800, which while not as speedy as eSATA, is still faster than USB 2.0.
Using the eSATA interface, the SmartStor DS4600 was the disputed performance winner on the ATTO test by a large margin. It also bears mentioning that the SmartStor DS4600 only suffered a small performance penalty going from RAID 0 mode to RAID 10 mode. On the other hand, the Vantec NexStar MX saw a noticeable drop in write performance going from RAID 0 mode to it's more robust RAID 1 setting, though its read speed stayed steady in both RAID modes. We also saw a write speed performance drop with the Drobo S when going from single-drive protection mode to Dual Disk Redundancy mode--the Drobo S's write speed of 58,996KB/Sec when using dual disk protection was the slowest of the group when using eSATA. Whether using single- or dual-disk protection, the Drobo S also had the slowest eSATA read speeds of the bunch--although, curiously we actually saw faster read speeds from the Drobo S using dual-disk protection than with single-disk protection.
The results played out similarly when using the USB 2.0 interface, with the SmartStor DS4600 still taking the read and write performance leads--although this time, the Drobo S was not far behind the DS4600. You'll also notice that the SmartStor DS4600 RAID 0 and RAID 10 scores are virtually identical--and the same can be said for the Drobo S's single and dual-disk protection scores as well. This tells us that the USB 2.0 interface's limitations are kicking in before the full performance capabilities of the devices can be achieved. As we saw with the eSATA results, the NexStar MX puts in slower performance when going from RAID 0 to RAID 1 mode. When it comes to using USB 2.0, the NexStar MX has the slowest write and read performance.
CrystalDiskMark is a synthetic test that evaluates both sequential as well as random small and large file transfers. It does a nice job of providing a quick look at best and worst case scenarios with regard to hard drive performance--best case being large sequential transfers and worse case being small, random 4KB transfers. The charts below reflect only the best-case scenario, sequential reads and write.
As we saw with the ATTO benchmark, the SmartStor DS4600 has the fastest eSATA write and read speeds on the CrystalDiskMark test--and here as well, there is little difference between the device's RAID 0 and RAID 10 performance. In fact, the SmartStor DS4600's write performance is more than twice as fast as that of the Drobo S in Dual Disk Redundancy mode. The NexStar MX takes the middle ground with a drop in performance going from RAID 0 and RAID 1 modes. And once again, the Drobo S brings up the rear with its write speed of 51.5MB/sec when using dual-disk protection. As a point of comparison, the 2TB Seagate Barracuda XT generated a write score of 142.7MB/Sec and a read score of 149.9MB/Sec on the CrystalDiskMark test in our 2TB hard drive roundup.
When using the USB 2.0 interface there is a much tighter grouping with the performance of all the devices--indicating that the USB 2.0 interface once again is the primary limiting factor. The SmartStor DS4600 still managed to squeak ahead with the write and read performance of both of its RAID 0 and RAID 10 modes, but the Drobo S is right behind it. The NexStarMX brings up the rear of the USB performance, and there is a noticeable speed loss when it goes from RAID 0 to RAID 1 mode.
|File Transfer Tests|
Our last series of tests are what you might call more "crude measurements" in that we simply fired up our trusty stopwatch and measured the time it took to complete a drag-and-drop of a single large file or a bunch of smaller files from an internal drive in our test system to the external device being tested. Our internal drive was a 120GB OCZ Vertex Turbo 2.5-inch SSD, which was connected to our testbed system's IOGear GICe711S3 PCI Express card. With our Large File test, we copied a single 3.4GB ISO file between the 120GB OCZ Vertex Turbo 2.5-inch SSD and the DAS being tested. With our Small Files test, we copied a 533MB folder (made up of 90 JPGs ranging in size from 2.27MB to 4.38MB, and 78 MP3s ranging in size from 1.98MB to 4.35MB) between the OCZ SSD and the DAS being tested. Note that due to some unexplained anomalies with the USB 2.0 results with these tests, we chose to include only the eSATA results.
While these are our only true "real-world" tests--copying real files from and to the test drives--the results here need to be taken with more than a few grains of salt. The results really only apply for the specific file-transfer scenarios we explore. You might very well see different performance results depending on what kind of file transfers you are performing. It's also important to note that these tests don't give a reliable indication of the sort of performance you might see with data transfers related to running specific data-heavy applications.
On our large file tests, the status quo remains intact--the SmartStor DS4600's RAID 0 and RAID 10 write and read performance still leads the pack. The Drobo S's performance is significantly slower than the SmartStor DS4600, and the Drobo S still suffers a performance penalty with its write performance when going from single-drive protection mode to Dual Disk Redundancy. The NexStar MX's RAID 0 performance falls between that of the Promise DS4600 and the Drobo S's single-disk protection mode performance, but the NexStar MX takes a write and read performance hit when using RAID 1 mode.
On our small files transfer tests, things start to switch up a bit. On this test, the Drobo S actually takes the lead on write speeds with both its single- and dual disk-protection modes. This lead is by a such a small margin, however (less than a one-percent difference), that the write performance of both of the Drobo S's modes, the SmartStor DS4600 using RAID 0, and the NexStar MX using RAID 0, are all statistically the same. The NexStar MX using RAID 1 and the SmartStor DS4600 using RAID 10 are a bit slower, but only by a very narrow margin. Our takeaway here is that the playing field becomes very close to even when it comes to small files write performance.
The same, however, cannot be said for small files read performance. Once again, the DS4600 takes top honors with both its RAID 0 and RAID 10 performance--and again with similar speeds from the two RAID modes. The NexStar MX regains the middle ground, and the Drobo S again has the slowest showing--although this time around the Drobo S's dual-disk protection mode was oddly faster than its single-disk protection mode.
|Performance Summary and Conclusion|
Performance Summary: There should be no doubt that the Drobo S offers a more flexible data integrity alternative than a traditional RAID DAS solution. But that flexibility gets you only so far if data transfer speeds aren't up to muster. The Drobo S consistently put out only mediocre data transfer performance, and also demonstrated a noticeable performance drop when set to its most aggressive fault tolerance setting. Depending on what you plan on using a DAS device for, its performance capabilities can be a mission critical factor. If you plan on using a DAS for a task such as non-linear video editing, then you're going to want to use a fast device, such as the Promise SmartStor DS4600 and connect it via the eSATA interface. On the other hand, however, if the device's primary purpose is for automated backups that take place in the background or during off hours, then performance shouldn't matter that much to you. Also, if you plan on using a slower interface connection, such as USB 2.0 or FireWire 400, then performance also becomes much less of an issue.
There are times, though, that the most important factor you want in a DAS device is data integrity, and the Drobo S shines in that department. You can start with as few as two drives and by default the Drobo S could lose one of the drives and still never miss a beat when it comes to uptime or maintaining data integrity--similar to RAID 5. If you have three or more drives, you could lose up to two of them and have a device that still holds all of your data and is still accessible--similar to RAID 6. But there is a lot more to the Drobo S than fault tolerance.
For starters, unlike RAID, you can use a mix of any size drives in the Drobo S. With a traditional RAID array, you either have to use the same-size drives, or the array will need to be configured based on the lowest-capacity drive of the group. With the Drobo S, you can add additional drives to an existing array and the new drives will automatically be configured to join the array and increase total storage capacity. You also swap out existing drives for higher-capacity drives, and the Drobo S will automatically integrate the new drives to the array.
What the Drobo S offers that traditional RAID-based DAS devices don't is a very flexible and seamless way to upgrade storage capacity and replace failed drives. But this level of innovation comes at a steep price. For the price of an unpopulated Drobo S, you could buy two Promise SmartStor DS4600 devices, plus a Vantec NexStar MX DAS, and still have money left over. You have to decide for yourself if the ease of use and flexibility of the Drobo S is untimely worth the significant price premium.