Best way to test network and disk read/write performance?

[heckofagator]

Hello group.

I have had a TS451 with the stock 1GB RAM for many years and have always been unhappy with its performance. Transferring files in and out and when using PLEX always takes longer than it should, I feel.

Today, I'm on a Win10 laptop, connected to a switch at 1GB and an trying to transfer some large ~1GB files. I am averaging maybe 1 MB/s transfer, as reported by Windows Explorer. The transfer will often drop well below 1 MB/s for long periods and will occassionally burst up to 6-7 MB/s every once in a while. I am using WD Green drives on the QNAP (I'm not sure if that matters). Transferring a group of files totaling around 5GB took me upwards of 20-30 minutes. I'd like to see if there is a way to test my environment and see where the bottleneck is.

Right now, during a file transfer, the QNAP Resource monitor is reporting an average of 10% average CPU usage and 560MB out of 879NB of the RAM.

[Moogle Stiltzkin]

hi,

in my faq i listed stephs perf qpkg as a way to test the network performance.

If you got MAC black magic may be a good way to test read/write perhaps? if not Intel has a NAS test app.

[heckofagator]

oh great, I will definitely check that out. And yes, I do have a Mac Mini so will check out Black Magic. thanks!

[Trexx]

Hello group.

I have had a TS451 with the stock 1GB RAM for many years and have always been unhappy with its performance. Transferring files in and out and when using PLEX always takes longer than it should, I feel.

Today, I'm on a Win10 laptop, connected to a switch at 1GB and an trying to transfer some large ~1GB files. I am averaging maybe 1 MB/s transfer, as reported by Windows Explorer. The transfer will often drop well below 1 MB/s for long periods and will occassionally burst up to 6-7 MB/s every once in a while. I am using WD Green drives on the QNAP (I'm not sure if that matters). Transferring a group of files totaling around 5GB took me upwards of 20-30 minutes. I'd like to see if there is a way to test my environment and see where the bottleneck is.

Right now, during a file transfer, the QNAP Resource monitor is reporting an average of 10% average CPU usage and 560MB out of 879NB of the RAM.

Would need more information (QTS version & build, NW configuration, storage configuration, etc.) but to start off with always good to rule out bad cables, etc.

SSH into your box, and do an ifconfig.

You should see something like this:

Code: Select all

eth0      Link encap:Ethernet  HWaddr 24:5E:BE:0B:B5:4C
          UP BROADCAST NOTRAILERS RUNNING SLAVE MULTICAST  MTU:1500  Metric:1
          RX packets:204171723 errors:0 dropped:5 overruns:122 frame:0
          TX packets:232306008 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:1000
          RX bytes:164575450927 (153.2 GiB)  TX bytes:124344610952 (115.8 GiB)
          Memory:fe300000-fe31ffff

Look specifically at the ERRORS: values, those should be generally 0.

Another thing I would recommend is to install the QNAP Diagnostics Tool, and then run an HDD Analyzer / Test Performance.

That will help you see what your individual drive throughputs are looking like at the HW level itself.

[ianmacd]

My only issue with black magic is it can only have maximum 5GB test blocks so actually doesn't show up the issue I am having where is doesn't kick in until after about 6-8GB. RFE into them I think to allow more options there.

[heckofagator]

I think this looks good? I'll go to look for the QNAP diag tool now

Code: Select all

eth0      Link encap:Ethernet  HWaddr 00:08:9B:E9:8D:F0
          inet addr:10.10.10.3  Bcast:10.10.10.255  Mask:255.255.255.0
          inet6 addr: fe80::208:9bff:fee9:8df0/64 Scope:Link
          UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
          RX packets:4169888 errors:0 dropped:0 overruns:0 frame:0
          TX packets:5740809 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:1000
          RX bytes:4732817883 (4.4 GiB)  TX bytes:8254764791 (7.6 GiB)
          Memory:90700000-9077ffff

[heckofagator]

From the diag tool

File System Analyzer

Code: Select all

Test volume : /share/CACHEDEV1_DATA
Block device : /dev/mapper/cachedev1
Left space : 1091952 MB
Test start : Mon Jul 17 10:45:52 EDT 2017
Test finish : Mon Jul 17 11:14:48 EDT 2017
Write 2GB performance : 2.36 MB/s
Read 2GB performance : 3.27 MB/s
Use extents : 5

HDD Analyzer

Code: Select all

Name:  Disk 1
Path:  /dev/sda
Performance:   115.92 MB/sec
Name:  Disk 2
Path:  /dev/sdb
Performance:    95.25 MB/sec

[dolbyman]

Write 2GB performance : 2.36 MB/s
Read 2GB performance : 3.27 MB/s

That is by far lower than it should be .. if the RAID OK ?

do a

Code: Select all

cat /proc/mdstat

[heckofagator]

yes, that number caught my attention, too, but then the HDD disk performance seemed ok. So maybe its something with how I set up my disks? Could be, I had never done a raid setup before.

Code: Select all

# cat /proc/mdstat
Personalities : [linear] [raid0] [raid1] [raid10] [raid6] [raid5] [raid4] [multipath]
md1 : active raid1 sda3[0] sdb3[1]
      1455183616 blocks super 1.0 [2/2] [UU]

md256 : active raid1 sdb2[1] sda2[0]
      530112 blocks super 1.0 [2/2] [UU]
      bitmap: 0/1 pages [0KB], 65536KB chunk

md13 : active raid1 sda4[0] sdb4[1]
      458880 blocks super 1.0 [24/2] [UU______________________]
      bitmap: 1/1 pages [4KB], 65536KB chunk

md9 : active raid1 sda1[0] sdb1[1]
      530048 blocks super 1.0 [24/2] [UU______________________]
      bitmap: 1/1 pages [4KB], 65536KB chunk

unused devices: <none>

[dolbyman]

oh so the drives are in single disk config ? and only two drives in a 4 drive unit ?

anyways .. even single disks should be much faster (as that HDD analyzer states) ... green disks are very prone to excessive head parking and shortened lifetime .. but the I am out of ideas ....

as a last straw.. any funny smart values on the disks ?

[Trexx]

How do you have your storage configured? Storage Pool, Static Volume, Thin Volumes, etc? Do you have snapshot enabled?

And again - need QTS Version & Build, etc.

Also WD Green drives are likely part of your problem. They do NOT perform well in NAS devices nor are they designed for NAS level of wear and tear.

[heckofagator]

Yeah, the Green drives are old and left over from my HP Media Smart (which died and then I moved to the QNAP). Everyone was always recommending the Greens back then.

Here's some screenshots. Does anything here standout?

[Moogle Stiltzkin]

Yeah, the Green drives are old and left over from my HP Media Smart (which died and then I moved to the QNAP). Everyone was always recommending the Greens back then.

Here's some screenshots. Does anything here standout?

watdaheck

j/k but uh greens uh.... i don't remember them faring that well back then
[youtube=]Y63rbPs_xzY[/youtube]

more specifically

The green drives have 8 second auto head parking (WD calls it "Intellipark"), which can cause issues in a RAID, and shorten drive life by increasing the number of drive cycles. You can Google "WDIDLE3", which is a tool to allow you to disable or increase the auto park interval.

this was a flaw on the greens, also Wd's Idle would cause the drive to park when not in use, which if the computer needed data, it would have to spin the platters up and unpark the heads to read/write the data. Which would cause the PC to hang.. also if you used these in a RAID they would drop out because of the extended time needed to unpark and spin up.

but wd red pro nas drives are nice costly so i opted for a hgst deskstar nas (cheaper) which sits between that and a wd red in terms of performance.

[heckofagator]

yeah, thanks for the thoughts. I always knew the right thing to do was proably replace the green with the red but as a light user, this issue only seems to bug me every so often. I'll do some work on the NAS and get so frustrated with the transfer times, I'm ready to throw it out of the window, and then I'll go a month without needing to mess with it much, so its not as noticable.

I always figured the spin down and park with the greens wasn't an issue for me - I didn't care about a few extra seconds to get up and going. I figured once the drives were in use, like during a sustained data transfer, the red and green might be pretty similar so I thought maybe the greens were all THAT bad. But I'm far from a HDD architecture expert, so who knows...

but yeah, some of these dismal transfer rates, its absolutely terrible at times.

[Trexx]

Based on your screen shots, you may also have filesystem issues (which is why the check file system option is there).

I would also upgrade to the latest QTS build 238 as there are alot of bugs and security vulnerabilities fixed between 188 and now.