One of the NAS aspects that does't get much attention in online reviews is data recovery. We have been trying to address that starting with our review of the LG NAS N2A2. In this section, we will first see how RAID rebuild works.

In order to test RAID recovery, we created a iSCSI LUN and filled it up with data so that the NAS became 96 % full. We removed the drive in the second bay, and the administration web interface immediately reports entry into the degraded mode. In the degraded mode, the read and write performance have negligible penalty, and the data access suffers no issues even with the rebuild going on in the background.

The removed hard drive was formatted to NTFS in Windows and reinserted into the drive bay. The rebuild went without any issues. However, repeating this with the drive in the first bay didn't go well. The following two screenshots from the system logs show the encountered issue.

 

This problem has also been reported on the QNAP forums here. The only way to get past this was to either insert a raw drive into the system or make sure all the partitions were completely removed from the drive to be reinserted (no NTFS or exFAT partitions). We rebuilt the RAID-5 volume three times under different conditions. The table below indicates the time taken for the rebuild process to complete with different replacement drives.

RAID-5 Rebuild Durations
1 TB Samsung 7200rpm Drive, NAS 96% Full 04:03:15
2 TB Seagate 5900rpm Drive, NAS Empty 04:10:20
1 TB Samsung 7200rpm Drive, NAS Empty 04:01:34
.

It is evident that the rebuild takes around 4 hours irrespective of the drive being used for the rebuild.

The next aspect we tested was the one touch copy feature of the TS-659 Pro II. We used a OCZ Enyo USB 3.0 SSD filled with 100 GB of data (multiple copies of the 10.7 GB Blu-ray folder structure used in the robocopy tests) and used the one touch copy feature to copy it into the RAID-5 volume. The transfer was carried out at 87.28 MBps.The USB 3.0 feature is a nice addition to the unit, but it only makes sense in situations where you have data on SSDs to back up. We can't imagine too many people using HDDs with USB 3.0 (where eSATA is available for cheaper). We would also have liked a one-touch copy feature for one of the eSATA ports.

The table below records the power consumption of the unit under various conditions with the 6 x 1 TB drives attached. The PSU is not terribly efficient, consuming more than 1W even when the unit is fully powered off.

TS-659 Pro II Power Consumption
Powered off 1.3 W
Idle (Sleep after no hard disk access for 30 min.) 30.2 W
Powered on 61.5 W
Powered on (Full load) 72.3 W
.

 

SMB, iSCSI and NFS Performance Conclusions
Comments Locked

69 Comments

View All Comments

  • Toadster - Monday, September 19, 2011 - link

    the specs on page #1 show 43W max power, but in the SMB, ISCSI and NFS page - you show 72.3W usage - which is it?

    also - why the heck are they using a 350W PSU when only 72.3W max? I could see maybe a 100W PSU which may remove the need for the PSU fan (thereby reducing noise?)

    overall, very tempted to get this device!
  • ganeshts - Monday, September 19, 2011 - link

    That spec on page #1 is QNAP's claims (and that is with 500 GB hard drives -- mentioned somewhere else). We measured 72.3W and stand by it :)

    Echo your sentiments on the PSU.. And QNAP claims 350 W PSU when the internal PSU is just 250W (not that it matters when the max power consumption is around 70 W only)
  • MichaelD - Monday, September 19, 2011 - link

    It's sad that a state-of-the-art, $1K+, SMB NAS device still is unreliable when it comes to rebuilds. I stopped using my D-Link DNS-343 (not in the same class as the QNAP) because of all the issues I had with it. Dog slow access speeds, lockups, you name. FW flashes fixed nothing.

    Granted, the web interface is very attractive and it has a lot of high-end features (and the display is nice looking) but when it comes down to brass tacks, data integrity and availability are all that matters. If I can't rely on this device to successfully rebuild after a drive failure, what good is it?

    I built my own NAS (server) out of a mATX mobo/RAM/CPU/HW RAID card I had laying around. I own a copy of Server 2003. I also owned the 5-drive SATA enclosure. Been up for almost 18 months now with a reboot roughly monthly for Windows updates. Zero issues.

    These "shoebox NASs" just aren't ready for prime time. STILL.
  • saiga6360 - Monday, September 19, 2011 - link

    Windows updates? LOL
  • jimr1234567890 - Monday, September 19, 2011 - link

    What are the effects of going from 1 GB to 3 GB memory? Any better performance? Faster rebuild time? Is there any benefit and if so what are they?

    BTW: like the article I like the fact you took out a drive and made the system rebuild a disk. Most articles I have read just gloss over any real world test and just regurgitate the products propaganda.

    I wonder how well this would work as source of video files from my DNLA complainant TV?
  • ganeshts - Monday, September 19, 2011 - link

    uPnP media server works fine as a source for your DLNA TV.. But, this one doesn't do transcoding.. so hopefully your TV's DLNA profile is good enough for your videos.
  • jimr1234567890 - Monday, September 19, 2011 - link

    What would be a good one that can effectively handle transcoding then?
  • ganeshts - Monday, September 19, 2011 - link

    I am evaluating something in the ReadyNAS lineup which has Orb inbuilt. This is supposed to have a transcoding engine, but I am yet to test its effectiveness.
  • saiga6360 - Monday, September 19, 2011 - link

    Why transcode at all? Get a proper media player. TV media players are crap.
  • DanNeely - Monday, September 19, 2011 - link

    Even if the box on your TV has enough computational power to play arbitrary modern formats without special purpose hardware low power handheld devices (phones, and tablets) are unlikely to be able to do so any time soon. Over longer terms unless we eventually reach a point where throwing more hardware at the problem stops yielding better quality images for a given file size I don't expect this to change.

    Also the number of playback points you have is also a factor. If you only need to stream to one or two TVs spending an extra hundred bucks per screen for more powerful decode isn't a big deal; if your McMansion has a tv in each of a dozen+ rooms then consolidating all of your transcoding into a single location and using dumb hardware at each TV will save significant amounts of money.

Log in

Don't have an account? Sign up now