|EonNAS 850X NAS Network Storage Server|
|Reviews - Featured Reviews: Network|
|Written by Bruce Normann|
|Thursday, 14 February 2013|
Page 11 of 15
1GB RAID 5 Test Results
If you've got more than three HDD spindle to put in play, it makes sense to use one of the more advanced RAID configurations. RAID 5 is one of the most popular setups, primarily due to the balance it exhibits between capacity and redundancy. Not surprisingly, most NAS units that can support more than three HDDs also support RAID 5, so it makes sense to use it for test purposes. Most NAS products that can support RAID 5 go beyond the minimum number of drive bays, to a total of four, so that is the number of drives that I typically use to test with, even though I could get by with only three. The EonNAS 850X and the QNAP TS-879U-RP have eight drive bays, which could offer increased performance over four-bay systems, but we want to test comparable configurations here. So, all systems are tested with the standard four disk contingent. We'll do a follow-up later, to measure 10GbE network performance, and then we'll load up all the available bays.
The results for RAID 5 Read testing show the EonNAS Pro 850X still slotted slightly below the high end QNAP and the Thecus N5550. All of these NAS platforms do a credible job here, though. None of them could be called a poor performer in a typical READ scenario; it's typically the Write performance that separates the men from the not-so-men. Lesser products typically have problems handling the extra processing required to manage all the parity and data striping machinations that go on with higher-value RAID configurations. None of the NAS products in this test suffer any real performance hits when they have to handle four drives simultaneously, instead of just one.
The 1 GB RAID 5 disk Read and Write tests show more clearly the kind of performance that a typical user would experience with these high-end NAS products. The EonNAS Pro 850X once more shows off its strong write performance, sitting right alongside the top-performing models from QNAP and Thecus. This is great news for anyone who is interested in deploying a NAS with the ZFS file system. This proves that with a reasonable amount of CPU power and a significant upgrade in the memory size, results can be obtained that are on par with Linux-based systems. RAID 5 Write performance can be a weak point in some systems, with all the computation overhead involved and the extra parity bits that need to be calculated and written to each of the drives. The only way to overcome that is with raw computational horsepower, which is why the ARM-based NAS units lag far behind both the Intel Atom and Core i3 units. It's an inescapable fact that the simplest assignment any NAS can perform is basic backup duty, and in order to do that task well, you need to buy the most powerful system to effectively reap the benefits of a multi-disk array. Don't scrimp on the NAS platform if you can help it.
Next up is 10 GB (1000 metric megabytes / 10,000,000,000 bytes) file transfer testing. Using the 4-disk RAID 5 configuration in each NAS, and a single Gigabit connection, network throughput will be put to the test, and the effect of any system or hardware caches will be minimized.
10GB RAID 5 Test Results
Looking at Read tests with a single 10GB file, the EonNAS 850X continues to show that Write performance is its strong point, not reading. That's not necessarily a bad thing, especially for a device that's going to be doing a lot of backup duty. The results still favor the more expensive models, even though it's not a 1:1 ratio of improvement with higher cost. It's instructive to compare the performance of the two EonNAS models, since they are the only two in this test using the ZFS file system. It's clear that the higher processing power of the i3-2120 CPU and the 8GB or RAM make a big difference in performance. I watched the CPU usage chart during the trials, and without deduplication and encryption enabled, the CPU never moved above the 25% load mark on Read tests, so there's plenty of reserve available. In order to do substantially better than this, you have to upgrade the network connection; GbE is only good for 125 MB/s, on a theoretical basis.
Looking at Write tests with a single 10GB file, the results are not all that different from the 1 GB tests. The Thecus N5550 runs at almost exactly the same average speed, in fact the Jumbo Frame results are identical. The same goes for the QNAP TS-879U-RP; the 9000 MTU write speed is the same, whether handling 1 GB files or 10 GB ones. The EonNAS Pro 850X also has basically the same average write speed with 10GB files, all within 1 GB/s. All three of the top performers just grin at the additional load of four hard drives; the CPU utilization never stayed above 25% during this Write test. The Intel Atom devices have to work a lot harder; the CPU was often maxed out during RAID 5 testing. The Marvell-based units always had the CPU maxed out in Write activity, and it really hurts their RAID performance.
As we wrap up the simple file transfer testing, it's pretty clear that the EonNAS Pro 850X is better at writing large files to disk, than it is reading them. It's also clear that Infortrend have tackled the ZFS performance penalty head on, and eliminated it as a potential problem for users. While additional memory was rather pointless in most of the Linux-based NAS units we've tested, the Solaris-based models experience real performance gains from bumping up the DRAM count.
We may see different results as we look at other tests, because these 1GB and 10GB files used in this portion of testing are much larger and more highly compressed than what a typical SOHO would feed their NAS. That's one reason Benchmark Reviews has expanded our testing protocol for NAS systems, to use a wider range of test data that's already in common use. Let's take a look at one of those tests, developed by Intel specifically for testing NAS devices. Oddly enough, it's called the "Network Attached Storage Performance Test"; NASPT for short.
NAS Comparison Products