Thecus N3200PRO NAS Server ReviewNemo - June 14, 2009
» Discuss this article (1)
RAID 1/5 Data Protection
To me, having a massive amount of storage for all your files is no good if you don't have a way to make sure the data is safe. One way to help prevent data loss due to a hard failure is through the use of a hard drive array that offers some form of redundancy such as a RAID 1 or RAID 5 configuration. Let's make one thing very clear from the start, a RAID array is not a substitute for a proper data backup procedure performed on a regular basis. However, it can help prevent data loss while keeping your data accessible even if a drive fails. To me this is one of the biggest benefits of a NAS server next to the extra storage space it can offer.
To see how well the N3200PRO handled the loss of a drive, I simulated a failure in both RAID 1 and RAID 5 setups and tested each array for data safety and the ability to rebuild the array once the 'faulty' drive was replaced. The N3200PRO supports hot swapping which means drives can be removed and exchanged without powering the unit down, leaving your data accessible the whole time. For both test, I simulated the 'failure' by removing a drive from the unit was it was still powered on.
For the RAID 1 test, I first built a two-drive RAID 1 array using disks 1 and 2 and designated disk 3 as a hot spare. Having a hot spare means it can be used by the system to automatically begin the process of rebuilding the array without the user having manually replace the drive. While the unit was running I removed HDD1 from the enclosure. With the hard drive removed, the unit responded with a warning tone and the HDD1 activity light turned a solid red indicating a problem. The LCD display on the front panel immediately displayed a “Disk Status Disk 1 removed” message and the unit also sent out a notification e-mail telling me the drive had been removed. When I checked the RAID screen in the management UI it showed the unit was already in the process of recovering the array using disk 3. Once the process was complete the RAID 1 array now consisted of drives 2 and 3. Once I returned drive 1 to the enclosure, the HHD1 activity light went back to normal and the system showed it as available to be designated as the new spare drive. I checked the data on the volume both during the rebuild process and after it was complete and found all the files were intact.
The RAID 5 test was similar, however as RAID 5 arrays require a minimum of three drives, there was no hot spare to automatically begin the rebuild process. I removed disk 3 and received the same set of warning messages via the front panel and through e-mail and the management UI showed the volume as degraded. Once I returned the disk to the enclosure, it showed up as a spare drive and the system then began the long process of automatically rebuilding the array.
I want to briefly touch on the capabilities of the N3200PRO to act as an FTP server. Once I had created a test user account and assigned the appropriate access rights I forwarded port 21, the standard FTP port, to the N3200PRO using its static IP address. I tested the functionality using the FireFTP add-in for Firefox and was able to connect with no problems. Response from the N3200PRO was extremely fast and the directory structure the test account had access to populated very quickly. I was able to transfer a large number of files (4,000+) and the N3200PRO displayed no problems and maintained a solid connection throughout, something I have had problems with on other servers. The N3200PRO performed flawlessly and I was surprised how easy it was to use. The only hiccup I encountered was when I mistakenly tried to send over an .iso image file that was larger than 2GB, which is the limit the N3200PRO will accept. Other than that I found FTP to be easier to use than Windows Explorer when moving files to and from the N3200PRO.
During the review process I checked out all the other features offered by the N3200PRO and covered most of the results during the configuration portion of the review. I didn't test the Web Camera or the UPnP multimedia features, but everything else worked as advertised aside from any problems I've noted previously.
I would like to address a couple of items I think Thecus needs to correct concerning the volume creation under the RAID section. First of all is the lack of support of standard, or single disks, when creating volumes. Every other NAS unit we've reviewed to date has this feature and Thecus should take note and add this functionality in a future version of the firmware.
Another issue I found in the RAID setup was the fact that the system would let you chose three drives for a RAID 1 setup and even create the volume using all three drives. RAID 1 volumes can only consist of two drives – a primary drive and a mirror drive. Creating a RAID 1 with three disks gives you the same space of a two-drive RAID 1 array which is what you expect, even though the Thecus shows all three drives as part of the array. I suspect the system automatically designated the third drive as a spare. To test my theory, I repeated the 'failed' drive test detailed on the previous page and when I removed one of the hard drives from the enclosure, the system showed the array as degraded and began to rebuild it. After the rebuild was complete and I reinstalled the drive in the server it automatically marked it as a spare. So you can build a three-drive array and the system automatically allocates the third drive as a spare even though the Spare checkbox was not ticked. This could be confusing to a novice/home user and just one of the touches Thecus should fix this to prevent confusion on the user's part.
Normally when a NAS appliance supports jumbo frames we run each test with and without jumbo frames enabled. In the past we've always used 9000 byte MTU frames which is the standard maximum size and the maximum size supported by our test rig. You'll notice we did not publish jumbo frame test results for the N3200PRO. The unit doesn't support 9000 byte MTU size, but it does support 4k, 8k, 12k and 16k byte MTU sizes. The closest match between the N3200PRO and our test setup was 4000 on the N3200PRO and 4088 on the test rig.
I ran each series of tests with and without jumbo frames enabled. The results were not sufficiently different to warrant publishing, especially since we couldn't directly compare them to the 9000 byte MTU sizes used by the other review units. In fact, for the N3200PRO the majority of the results were 1-6% worse with the remaining few cases being the same or about 1% better, which is not statistically significant.
The Intel Gigabit Ethernet controller can support jumbo frames up to 16k as implemented on the N3200PRO, what I don't understand is why Thecus chose to implement the ranges they did as most home equipment won't support greater than 9000 byte MTU sizes. Perhaps I'm belaboring a minor point as home users are likely to enable jumbo frames as all other devices on the network must be running a a Gigabit connection and bet set to the identical MTU size.
These types of problems are not fatal flaws in the N3200PRO however and are more a reflection on the lack of maturity in the firmware. The N3200PRO has a lot of good features and strong performance for its class and hopefully Thecus will continue to improve the firmware and implement fixes for these types of issues in the future.