NAS System Overhead Measurements

I've discussed the potential impact the NAS hardware has on performance in general terms so far. The hard reality is that the CPU, drive controllers, memory, and network subsystems have a direct and profound impact on the throughput of a NAS device. In extreme cases where multiple drives (4+) are arranged in higher-order RAID configurations, the CPU has a ton of work to do, calculating parity bits and parsing them out to multiple data streams. In-line data encryption adds another potential load to the infrastructure, but I'm not going to bother testing that until I get a system with the recent AES-NI additions to the Intel64 instruction set. It's just not realistic to use encryption without that level of support. In this section, I'm going to look at some results from the System Monitor capability that is available on the QNAP Turbo NAS server.

Let's start off looking at CPU usage on the NAS server. During a straight data transfer from the PC to 4 disks configured as RAID 5 on the TS-419P II, the results show the Marvell 88F6282 SoC device completely maxed out at 100%. The two blocks in the chart represent two different sets of files being transferred, with a break between them. The first block represents ten 1GB files transferred by one Windows command, and you can see little dips where the CPU paused between each file. The second block represents one 10GB file, transferred the same way. Any dips in CPU utilization in this section are due to system wait states thrown in by some sub-system crying "Uncle" for a brief period of time. By and large, there's little difference between the two scenarios, except the single 10GB file is transferred over a little quicker.

The memory subsystem on the QNAP TS-419P II is not being taxed by these file transfers at all. Unless you plan to use the NAS for all of the "extra" things it can do, as a media server and such, don't worry about the fact that it only comes with 512MB of memory capacity. That's plenty, at least for the basic disk functions. This is very close to the same level of memory usage I saw on the two-bay device I tested , so there's no additional load presented to the memory by four disks in RAID 5, as compared to the single-disk configuration.

The network interface is getting more of a workout than the memory, but it is still running well below the throughput limits of the Gigabit Ethernet (GbE) interface. This particular chart was produced during two disk-writing tests, we'll see a difference with disk-read tests shortly. There's a lot of extra capacity here, even with one of the larger NAS devices, running multiple disks in RAID configurations with a single GbE connection. This TS-419P II unit, and most of the larger QNAP units, all have dual Ethernet connections that allow for teaming via IEEE 802.3ad/Link Aggregation, that in certain cases allows for almost double the network throughput. This is only going to be required in rare cases, where both systems connected this way have the raw transfer speed to make it necessary. That's the sort of thing you're only going to see in a corporate LAN room, at least for now. One day, I'm going to load up one of the big NAS units with high end SSDs in RAID 0 and let it rip; then we'll see where the system bottlenecks are.

The network throughput scales right along with the disk throughput, as seen here. In this test, I was reading from the NAS, instead of writing files to it. The increased data transfer rate from the disks translates to an increase in network throughput, up from about 55MB/s in write operations, to over 80MB/s in reading mode. In the following chart, you can also see the effect on the network load as a multi-file transfer operation takes place. Until now, we have been moving very large files around, but in the last portion of the chart below, I transferred a very large number of smaller files, just to see how that would impact performance. Instead of one 10 GB file, the green portion of the chart shows what happens when transferring 4,793 items totaling 6.5GB from the PC to the NAS. If you're going to use any NAS for basic backup duties, this is the kind of action it's going to see. Still, there should be no worries as far as network speed goes, no matter what the primary usage is.

In contrast, the NAS CPU is still being taxed during these file transfers, with either small numbers of large files or a large number of small files. In the third section of the chart below, you can see some additional dips, but they're sharp drops with a corresponding sharp recovery. There's nowhere for the CPU to hide in a high performance NAS appliance, and the ARM processor in QNAP's lower-priced models gets hammered pretty bad in typical use cases.

I hope this section showed you some objective reasons why the infrastructure that any NAS product brings to the table is important to its overall performance. As the number of drive bays goes up, the hardware requirements increase as well, and the price has to follow. I know it's disheartening to see that you don't get great economies of scale on the larger NAS units, but it would be even more of a shame if they didn't perform up to their true capabilities because the hardware was holding them back.

Now, let's look at some Final Thoughts, and then move on to our Conclusion and Product Ratings.

<< Previous Page - Next Page >>