Post your performance results

[sfg]

Bernhard Hartl wrote:

sfg wrote: I recently received my readynas pro business and while I am liking it I have not been getting anywhere near the speeds as it seems are possible. I am getting drag and drop read and write speeds of about 4-6MB/s. I have been trying the various stuff reported here but nothing is seeming to cause much difference. The only way to get the full performance of the readynas pro is to connect via GigaBit LAN You have to use a good GBit Switch and CAT6 cables for connecting the computers and the NAS to the switch WLAN is always slow - that's not a problem of your NAS but a problem of WLAN

Even when connecting at full N speeds?

I was hoping for 20s or so even using wlan.

[starg4ze]

sfg wrote:

Bernhard Hartl wrote: sfg wrote: I recently received my readynas pro business and while I am liking it I have not been getting anywhere near the speeds as it seems are possible. I am getting drag and drop read and write speeds of about 4-6MB/s. I have been trying the various stuff reported here but nothing is seeming to cause much difference. The only way to get the full performance of the readynas pro is to connect via GigaBit LAN You have to use a good GBit Switch and CAT6 cables for connecting the computers and the NAS to the switch WLAN is always slow - that's not a problem of your NAS but a problem of WLAN Even when connecting at full N speeds? I was hoping for 20s or so even using wlan.

I manage to get 8-9MB/s over AFP using my MacBook Pro and a Timecapsule (both supporting 803.11n). You you won't see much better figures unless you are in an isolated area with no interference from other WLANs around you(or microwave ovens...) and sitting next to your AP with your computer.

[emonkia]

I am easily seeing write speeds of 50MB/sec. I have journaling on, and I am not using Jumbo frames. I think with a small amount of optimization, I could max out the NICs or the CPUs on the ReadyNAS, but either should be considered "really good" speed.

Sample rsync transfer of a few compressed LUN copies (HDD images):

Number of files: 48
Number of files transferred: 25
Total file size: 21878830041 bytes
Total transferred file size: 21878829859 bytes
Literal data: 21878829923 bytes
Matched data: 0 bytes
File list size: 1130
File list generation time: 0.028 seconds
File list transfer time: 0.000 seconds
Total bytes sent: 21881503093
Total bytes received: 704

sent 21881503093 bytes  received 704 bytes  55607379.41 bytes/sec
total size is 21878830041  speedup is 1.00
---- 2009-03-19 19:21:40 finish rsync

rsync server:
ReadyNAS Pro
HDD - 6x 1.5 TB Seagate (write cache on)
RAM - 4 GB
NET - 2x 1 Gb 802.3ad LACP bonded NICs
Journaling: on
Jumbo frames: off

switches:
Cisco ethernet - 3750G
Cisco fiber - 9124

rsync client:
IBM HS20 8843 Blade server
CPU - 2x 3.0 GHz Intel Zeon (HT)
RAM - 4 GB
HDD - SAN - LUN Raid5(3) via QLogic 2312 1x 2 Gb Fiber to Xiotech Mag3D 48 drive (non-idle)
OS - Ubuntu v6.06
rsync compression was off, not using SSH

I am easily able to pull 150 MB/sec with this machine, and depending on the current SAN load it can peg at around 200 MB/sec (SAN has faster IO than server's 2 Gb Fiber connection). The above was done while SAN was under normal load, with no configuration changes done to optimize speeds (standard everyday setup). I also transferred a much larger data set than the RAM in the ReadyNAS, so full caching was not a factor for speed shown.

This Pro just flat out blows the doors off the NV+ (as expected). If you need speed, get the Pro.

[rstone]

I guess I'll add my results as well...

ReadyNAS Pro Business Edition:
3072 MB RAM
6 x 1000 GB WD10EACS
All journaling disabled
Fast writes on
Jumbo Frames On (9000)
Dual Gigabit (Teamed using 802.3ad w/LACP)

Switch:
HP Procurve 1800

Test Machine:
EVGA X58 w/Intel Core i7 at 2.7Mhz
6 GB Corsair Memory at 1600Mhz
2x WD Raptors (RAID 0)
Vista x64
Intel Pro/1000 Dual PT NIC w/ Jumbo Frames Enabled at 9k

IOMeter Results:
Test: 256KB Sequential Read / 256KB Sequential Write
Read: 65.16 MB/s
Write: 94.35 MB/s

Real World Results: (Using copy & paste w/5 GB file)
Read: 109 MB/s
Write: 110 MB/s

[Lithium-Admin]

I ran some speed tests while I was trying to figure out all the crap about MTUs. I just copied and pasted from my other post (viewtopic.php?f=28&t=26902).

Hi Everyone!

Since no one responded I just went ahead with some trials. No offense taken because the post was LONG! :lol:

Before I give my test results here is my set up:
MBP (early 08, mac os x 10.5.6)
D-LINK DIR-655 Wireless Router (supports jumbo frame-9000, from my extensive research and Gigabit LAN ports, but MTU value is set @ 1492)
D-Link DGS-2205 (Gigabit Switch, also supports jumbo frame-9000, http://www.newegg.com/Product/Product.a ... 6833127083)
ReadyNAS Duo (MTU 7936)
Hostname: ReadyNAS
Model: ReadyNAS Duo [X-RAID]
Serial:
Firmware: RAIDiator 4.1.4 [1.00a042]
Memory: 256 MB [2.5-3-3-7]
IP address: 192.168.0.199
Volume C: Online, X-RAID, 2 disks, 76% of 925 GB used

All day I've been reading other people's post in regards to performance Mac/PC, ReadyNAS Duo/PRO... I've been messing around with the MTUs and file sizes. Here we go.
I did my best to get the numbers.

Note: The way I changed my MTU values were via Terminal by using the "sudo ifconfig en0 mtu" command. I double, triple checked MTU values before, after and during the file transfers. All of the file transfers were drag & drop over AFP.

Test 1
MBP MTU: 1500
NAS MTU: 7936
File Size: 4.25 GB
Read (from NAS to MBP): Average 42.1 MB/s, Peak 43.0 MB/s, Range 34 MB/s to 42 MB/s
Write (from MBP to NAS): Average 9.5 MB/s, Peak 11.1 MB/s, Range 9 MB/s to 11 MB/s

Test 2
MBP MTU: 7936
NAS MTU: 7936
File Size: 4 GB
Read (from NAS to MBP): Range 36 MB/s to 37 MB/s
Write (from MBP to NAS): Range 9.5 MB/s to 10 MB/s

I thought this was really odd because there was no change in any of the transfer speeds. So, I unmounted the drives and remounted it and that's when the transfer speed kicked up!

Test 3
MBP MTU: 7936
NAS MTU: 7936
File Size: 8 GB
Read (from NAS to MBP): Peak 52.3 MB/s, Range 49 MB/s to 51 MB/s
Write (from MBP to NAS): Peak 41.1 MB/s, Range 38.3 MB/s to 40 MB/s

Nice!!! A huge increase, but I think the read speed got a big jump because the file size changed to 8 GB. Huge improvement over the write speed.

This time I wanted to restart the MBP and NAS because I wanted to clear any cache (just my guess...). So I restarted both the NAS and MBP and changed the MTU on the MBP to 9000 to see what would happen.

Test 4
MBP MTU: 9000
NAS MTU: 7936
File Size: 4 GB
Read (from NAS to MBP): Average 35.5 MB/s, Peak 41.5 MB/s
Write (from MBP to NAS): Average 40.4 MB/s, Peak 41.3 MB/s

The read performance had decreased (from 50s to 40s) so I assumed that it was because the MTU value on the MBP didn't match the NAS MTU. So I reverted back the MBP's MTU to 7936 and restarted only the NAS to clear any cache.

Test 5
MBP MTU: 7936
NAS MTU: 7936
File Size: 4 GB
Read (from NAS to MBP): Average 36.0 MB/s, Peak 43.9 MB/s
Write (from MBP to NAS): Average 40.2 MB/s

So basically, using MTU value of 9000 or 7936 on the MBP had not that much effect. No significant difference in the read speeds. But I tried copying a 8 GB file on the same setting from Test 5.

Test 6
MBP MTU: 7936
NAS MTU: 7936
File Size: 8 GB
Read (from NAS to MBP): Average 51.5 MB/s, Peak 53.2 MB/s
File Size: 4 GB
Read (from NAS to MBP): Range 45 MB/s to 50.8 MB/s
Write (from MBP to NAS): Average 39.4 MB/s


Test Conclusions:
1) It's my believe (guess) that the read speed when the MTU on the MBP (either 7936 or 9000) is largely depended on the size of the file. The bigger it is 4 GB> the faster the read speeds.
2) Write speed will significantly improve when you put the MBP's MTU to 7936 or 9000. I now always hit an average of 40 MB/s.
3) My guess is that you should always mount and remount the NAS whenever you change the MTU values on the MBP (see test 1, 2 and 3).

Still remaining questions :roll: :
1) Does using jumbo frames reduce performance when streaming videos from the NAS?
2) For some reasons the MTU values on the MBP do not stick. I still have to change it each time after restart. It will always return to default value of 1500. Anyone know how to make the new MTU value stick?

Thanks a lot everyone!

[dbott67]

I can't answer either of your questions, but I can provide a little info about the jumbo frames on the ReadyNAS. It is my understanding that the ReadyNAS really only utilizes jumbo frames when WRITING to the NAS. You will not see any performance gains while reading.

Performance does vary, as you've noticed. Some of the factors include: NAS settings (type of RAID, # of disks, journaling, etc), amount of RAM, file size, number of files, network protocol used and contention/congestion issues.

Your numbers are quite good and it looks like you've got it tweaked for maximum performance. Thanks for taking the time to post your findings.

-Dave

[Lithium-Admin]

@ Dave

Thanks for the reply Dave! On this post and my other post!!! Hahaha... yes... it was very sad because it definitely seemed that I was talking to myself. 😄

Much appreciated!

-Han

[fishgeekted]

ReadyNAS Duo - Factory Presets

Desktop 10/100 Windows XP SP 3
Laptop 10/100/1000 Wi/Fi B,G,N Windows Vista SP 1 64-bit Home Edition
Router D-Link DIR655 X-Treme N Gigabit

Iometer tests:
XP 10/100 Desktop Reads 9.27 MB/S and Writes 10.49 MB/S

Vista Wireless N Laptop (Wireless 130 mb/s connection) Reads 6.5 MB/s and Write 9.07 MB/s ( I have tried tweaking with many of the suggestions here and have only slowed the laptop)

I started doing these tests when I discovered the NAS was having a huge amount of TCP Retransmits and the laptop was timing out on accessing certain folders on the NAS...

Are these average speeds?

Thanks for your help, I'm still learning how to use the NAS!

[dbott67]

For the XP 10/100 connection the numbers are very good (100 mbps = 12.5 MB/s which is the theoretical max; normal is 70-80% of that).

For wireless N, your number are also very good. Most people using wireless G get 2-3 MB/s and up to 6 using wireless N, so you're above that range.

-Dave

[cpitchford]

I've yet to test my Pro Pioneer on a network, I really wanted it to provide services more than remote file systems.

I've just upgraded from the stock 1Gig (4-5-5-15 667MHz) to 4Gig (6-8-8-18 800MHz) and I've seen some surprising speed boosts with raw IO to the disks..

Using hdparm -t -T /dev/c/c

Before upgrading RAM:
Timing cached reads: 1890 MB in 2.00 seconds = 945.13 MB/sec
Timing buffered disk reads: 510 MB in 3.00 seconds = 170.11 MB/sec

After upgrade:
Timing cached reads: 2350 MB in 2.00 seconds = 1175.62 MB/sec
Timing buffered disk reads: 550 MB in 3.00 seconds = 183.07 MB/sec

That equates to a 7% improvement buffered read and a 24% improvement cached read!

It would be really interesting to see how this NAS performs with load balanced Ethernet tests and multiple hosts but (without seriously trivial hacking) this feature is limited to the business model (frankly I think that is insulting on a $1400 NAS.. QNAP don't charge extra for load balancing or iSCSI :neener: )

Needless to say, I was surprised (and pleased!!) that a ram upgrade produced such an improvement! Best $60 spent!

There were some other performance questions:

Why a 32bit kernel? Not sure if 64bit would be better, but for TB systems I'd have though it possibly might be?
Why no hyper-threading. The Core2 Duo chip has two cores both used by the system. Both cores support hyper-threading but the kernel lacks support (it is turned off) Why wouldn't hyper-threading be enabled. I appreciate it is a minimal performance boost, but still..

[Lithium-Admin]

Macbook 10.5.6 (No Hacks)
Duo 256 mb - 2 x500gb (original seagates) (hacked sysctl)
Tested with Net Monitor transferring 3.5gb dvd iso
Receive 41.2 MB/s Peak 43.5 MB/s
Transmit 37.2 MB/s Peak 38.1 MB/s
2 Tcp Retransmit errors

[emonkia]

Why a 32bit kernel? Not sure if 64bit would be better, but for TB systems I'd have though it possibly might be?

EM64T support could be a benefit, but I am not so sure a full 64 bit system would be faster. The other thought is, why not. With the RAM limit at 4 GB, the choice is not clear, but there is no worries about future 64 bit support.

The below is curious though:

NAS home page:
Memory:   	 4096 MB [4-5-5-18]

# cat /proc/meminfo 
MemTotal:      2852720 kB
HighTotal:     1964736 kB
LowTotal:       887984 kB
SwapTotal:     2097080 kB

# free
             total       used       free     shared    buffers     cached
Mem:       2852720    2801552      51168          0     541108    1833832
-/+ buffers/cache:     426612    2426108
Swap:      2097080        132    2096948

Why no hyper-threading. The Core2 Duo chip has two cores both used by the system. Both cores support hyper-threading

As far as I know, the Intel E2160 does not have hyperthreading. Intel verifies it is a Core 2 Duo dual core chip.
http://processorfinder.intel.com/detail ... Spec=SLA3H

This Intel link specifically shows that the E2160 does not have hyperthreading:
http://ark.intel.com/cpu.aspx?groupID=29739

[beisser1]

and not all core2 duos support hyperthreading.. i have a q6600 quadcore (which is based on 2 6000 series dualcores) and that one definitely doesnt support hyperthreading.

since the e2xxx series is based on the 6000 series (just with less cache) i would suspect that they also dont support HT.

[dashyt2dafulliz]

anyone know any other ways to test speeds? Seems I'm not authorize to view OP link.

[Jonavin]

kreischweide wrote:
My performance results with my ReadyNAS Pro Pioneer Edition using the mentioned default Iometer settings: Total I/Os per Second: 309.62 Total MBs per Seond: 77.41 Average I/O Response time (ms): 3.2287 Maximum I/O Response time (ms): 173.7743 Using 4x 1.5TB Seagtes, D-Link DGS-1008D Switch, Jumbo Frames enabled, Intel C2D E8200, 8GB RAM and Onboard-NIC. Most satisfied so far.

The DGS-1008D is listed as incompatible but reported in 06. Are the newer revs fixed?

[walshlink]

NAS: ReadyNAS Pro, 2x 1TB Seagate ST31000340NS and 2x 1TB WD WD10EADS HDDs in X-RAID2, 4GB RAM, RAIDiator 4.2.5, jumbo frames enabled, disk write cache enabled, full data journaling disabled, optimized for Mac OS X, AFP file protocol enabled, jumbo frame set at 9000, NICs in 802.3ad.

Switch: Netgear GS108T latest firmware, all ports flow-control disabled, jumbo frames enabled, NAS NICs in LAG and LACP enabled.

I am able to write a 9.98 GB file to an APF share on the NAS at 85 MB/s. Using ATTO iSCSI initiator (which is faster than the globalSAN initiator), I can write a 9.98 GB file to the NAS at 37 MB/s. I can write to my uMBP (with Intel X25-M 160GB SSD) at approximately 65 MB/s with either one (limited by the SSD).

[Richy_Boy]

Hmm, I'm not seeing the performance that some of you guys are on our ReadyNAs Business 6TB unit.

ReadyNAS PRO side:
- 3GB RAM
- Bog standard 6x 1TB drives - XRAID
- No journaling
- Paired gigabit ethernet

Server side:
- Supermicro server
- SATA HDD
- Single gigabit ethernet
- 2GB RAM

Switch:
HP Procurve gigabit switch
No jumbo frames enabled
CAT5e throughout...

When running a ReadyNAS backup copying from the server to the ReadyNAS over CIF I'm only seeing a 25% network utilisation (but rock solid 25%), which is odd. I haven't had time to look into why this is so low, but the Procurve isn't doing much, I'm not getting any errors, no high CPU utilisation and I'm assuming the source 'server side' SATA drive can rummage up data quicker than ~30MB/s.

The 25% utilisation is confirmed in both the Windows GUI and the Procurve port utilisation screen.

Data being copied is a bunch of installation files for a variety of applications of random sizes. i.e. some 700MB ZIP files, some 30MB installers...

Any ideas?

UPDATE: Doh, this does seem to be the SATA drive maxing out. I just copied another data stream from a second drive on the same server and network utilisation went up to 50%... i.e. 62MB/s...

It does seem a little odd that when copying from a single server source it seems to max out around 40% utilisation, i.e. 50MB/s though, unless I'm now maxing out 4x 10k rpm SAS drives in RAID5!

Rich

[GrZeCh]

Hello,

what do you think about this results:



This tests was runned on VMWare ESXi 4.0 virtual machine (Windows 2008 x64 with 2GB RAM). Whole Windows system is placed on ReadyNAS 2100. I'm using switch Allied Telesis AT-GS900/8 switch (http://www.alliedtelesyn.pl/products/de ... =57&lid=16) with enabled jumbo frames (it looks like this switch supports them because I've set jumbo frames on ESXi - http://blog.scottlowe.org/2008/04/22/es ... bo-frames/ - and ReadyNAS 2100 and results had improved).

ReadyNAS 2100 with X-RAID and journaling enabled

[sodablue]

NAS:
ReadyNas Duo with Raidar 4.1.5
Two Western Digital 500G drives in X-RAID
Two Western Digital 500G USB drives
1 Gig of RAM
APC Back-UPS RS 1500
All journaling turned off, CIFS fast writes turned on, USB fast writes turned on

Client:
Intel DP35DP with E6600(Core2Duo 2.4Ghz) and 8 Gigs RAM
Windows 7 RC x64

Connected through Netgear GS108 switch

Copying files I am getting around 20Meg/sec writes and 40Meg/sec reads to the X-RAID shares.

And around 10Meg/sec writes, 20Meg/sec reads to the USB shares.

[peteshoes]

I am very happy with my Wired Perf, but not my wireless on XP laptop:

Model: ReadyNAS Duo [X-RAID]
Firmware: RAIDiator 4.1.6 [1.00a043]
Memory: 1024 MB [2.5-3-3-7]

MTU = 1500

Hitachi HDT721010SLA360 [931 GB] x 2

Netgear DG834GT Router 4.1.6 firmware (4.1.9 no good with PS3!)

Netgear GS605 GBE Switch - just added 🙂

Using the Iometer tool and suggested settings:

READ
MBps
38.702358

WRITE
MBps
20.894208

Don't think I could gett much better 😄

A hint for others having problems - make sure you disconnect EVERYTHING else from the NAS before running the test, it made a massive difference appox 50% or more when I accidently left something connected to the nas, even thought it was not apprently doing anything!

My lappy speed on 54Mg connection is bad though

Read = 0.6meg
Write = 1.08Meg

Very disapointed and not sure how to improve without slowing my wired speed!!??

Any advice most welcome.....

[dbott67]

Try setting the MTU on the NAS to 1492 or 1460 in Frontview > Network > Interfaces.

[guiri]

Since I'm too stupid to do the iometer thing, can someone here just drag and drop some files from the computer to the nas and tell me how long it takes?

I move a 2.15gb file in just about 5 mins either way which I think is very slow. This is an avi file.

Can anyone comment?

Thanks

GEorge

[peteshoes]

Well - here are the results I got to when testing with the 1gig file that iometer produces:

1 gig file

MTU 1460
copy file to PC - 1 min 37
copy file to nas - 1min 5secs
12mb/s 22mb/s


MTU Jumbo
46 secs
1min 40 secs
30.2mb/s 13.8mb/s

MTU 1492
33 s
1min 9 secs
39mb/s 22.6mb/s

Pete

[jaggy27]

Wireless Performance:
Read: 3.26894MBps, (13.07576 IOPS)
Write: 2.220606MBps, (8.882425 IOPS)

Direct Connect:
Read: 16.317407MBos (65.269628)
Write: 20.365237MBps (81.460949 IOPs)

Router: D-Link, DL-4500
Updated to latest firmware

ReadyNAS Duo
1 Single stock 500GB drive

HP pavillion dv6000 Laptop
AMD Processor, Core 2 Duo
2GB ram
Windows 7 OS

MTU: 1492 on both Laptop and ReadyNAS Duuo

Radiator at Version 4.1.4
Disabled full data journaling
Disabled journaling
Enable fast CIFS writes
Enable USB disk writes

Can anybody help with some suggestions to better speed????

[Copernicus2]

Are people with Windows 7 having the same speed issues I am? Running an IO meter will max my transfer speeds at about 8 MB/s both write and read. Everything seems to be running at gigabit speeds between my computer to the nas. When I switch to XP sp3 I have no issues reaching 29 MB/s.

Is this a Windows 7 issues or what?

Readynas NV+ 4.1.6 firmware