× NETGEAR will be terminating ReadyCLOUD service by July 1st, 2023. For more details click here.
Orbi WiFi 7 RBE973
Reply

Re: Post your performance results

musicfan
Aspirant

results from direct-connect re-test

PentiumD 3GHz 2GbRAM, LinkSys BEF SR41v4 Etherfast CableRouter with Cat5 cable

NV+ 1024Mb RAM, 4 disks X-RAID 500Gb Seagate ST3500630AS, Radiator 3.01c1-p6, UPS monitored, enable cache, disable journal, fast CIFS, Speed Auto-Negotiation, Smart disk check without problems.

Averages of several standard IOMeter protocol runs with and without jumbo frames enabled on NV+ (no difference) yielded...

Mb/s Write=10.5
Mb/s Read==9.0


Thank you all. My quote should have said MB/s. It was suggested the Etherfast 10/100 switch may be the speed limit. And, I should bypass the switch with a direct-connect. Finally...

Jumbo off, enable cache, disable journaling, CIFS writes, on UPS. Run both standard and with the maximum disk size quadrupled to 8192000 sectors with only a small difference.

MB/s Write 22
MB/s Read 44

Does this diagnose the venerable LinkSys BEF SR41v4 Etherfast CableRouter 10/100 switch as the speed limiter or bottleneck? Thank you.
Message 76 of 309
bhoar
Aspirant

Re: results from direct-connect re-test

musicfan wrote:
Does this diagnose the venerable LinkSys BEF SR41v4 Etherfast CableRouter 10/100 switch as the speed limiter or bottleneck? Thank you.


Yes. It's not a deficiency of that particular switch, it's just that a 100 megabit switch can only move data at up to 100 megabits, which is what the original numbers were showing. If you want to get closer to the direct connect speeds, you need a 1000 megabit (aka gigabit) switch instead.

-brendan
Message 77 of 309
StrawDragon
Aspirant

Re: Post your performance results

I am using a Netgear GS108T 1Gigabit switch with a top notch quad core Intel setup running XP SP2. Jumbo frames is enabled.

READ: 31.98MB/s
WRITE: 24.91MB/s

Thought it might be a bit faster tho before I purchased it. But its ok, nothing special.

Also running X-RAID.
Message 78 of 309
glam
Aspirant

Re: Post your performance results

I connect to my NV+ using two different computers, and I noticed (with Retrospect) that they seemed to get very different performance. My main use of the NV+ is for backup using Retrospect.

I ran Iometer on both while directly connected to the NV+ (and separate times, of course). I used the tests specified at http://www.infrant.com/forum/viewtopic.php?t=265 .

The NV+ is configured to use jumbo frames. The disk write cache is enabled. Journaling is disabled.

Here are the results that I got:<ol><li>IBM ThinkPad T41 running WinXP Pro (SP2). The NIC is an Intel(R) PRO/1000 MT Mobile Connection with driver version 8.6.11.0 (6/29/2005). Jumbo frames are enabled at 16128 bytes (the highest setting offered). I got 7.285511 MBps for writes and 3.192311 MBps for reads. (The numbers without jumbo frames were 5.608709 and 3.170804, respectively.)<li>Dell Inspiron 600m running WinXP Home (SP2). The NIC is a Broadcom 570x with driver version 8.27.1 (8/31/2005). I did not see any configuration option for jumbo frames. I got 8.185093 MBps for writes and 26.862328 MBps for reads.</ol>Various questions come to mind:

Should I be getting better performance (for either or both computer)? If so, what should I be doing?

Is there something that I can do to improve the performance of the T41? Should try a different setting for jumbo frames (e.g., 9014 bytes)? Should I disable jumbo frames altogether? (I had noticed the original discrepancies when none of the computers, including the NV+, were using jumbo frames.)
Typically, the NV+ is connected directly to a NETGEAR GS105 switch. When wanting performance, the other computers are typically connected to the same switch or to a NETGEAR GS108 switch (which connects to the GS105). Both of these switches are gigabit, and I believe that they support jumbo frames. Is there anything that I would want to do to get best performance in the typical situation?
Message 79 of 309
Lithium-Admin
Novice

Re: Post your performance results

Hi there

Using a Core2Duo E4400 with a D-Link DGE-530T-GBIT-NIC and a Netgear GS108 GBIT Ethernet switch I get the following results:

Read: 29.598 MBps
Write: 12.243 MBps

Jumbo frames are enabled (9014), journaling is off, AV scanner disabled.

I am a bit worried about the slow write performance and at this moment I don't have a clue how to correct this.

I opened another thread at http://www.infrant.com/forum/viewtopic.php?t=12661 for my speed issue to keep this one clean.

Olove
Message 80 of 309
bhoar
Aspirant

Re: Post your performance results

olove - not sure if this is the reason, but please check the wiki for the maximum size of jumbo frames supported by the unit. It's < 8000.

-brendan
Message 81 of 309
musicfan
Aspirant

Re: Post your performance results


PentiumD 3GHz 2GbRAM, LinkSys BEF SR41v4 Etherfast CableRouter with Cat5 cable

NV+ 1024Mb RAM, 4 disks X-RAID 500Gb Seagate ST3500630AS, Radiator 3.01c1-p6, UPS monitored, enable cache, disable journal, fast CIFS, Speed Auto-Negotiation, Smart disk check without problems.

Averages of several standard IOMeter protocol runs with and without jumbo frames enabled on NV+ (no difference) yielded...

MB/s Write=10.5
MB/s Read==9.0

Direct-connect:

Jumbo off, enable cache, disable journaling, CIFS writes, on UPS. Run both standard and with the maximum disk size quadrupled to 8192000 sectors with only a small difference.

MB/s Write 22
MB/s Read 44
--------------------------------------------------------------------------------

musicfan wrote:
Does this diagnose the venerable LinkSys BEF SR41v4 Etherfast CableRouter 10/100 switch as the speed limiter or bottleneck? Thank you.

brendan replied:
Yes. It's not a deficiency of that particular switch, it's just that a 100 megabit switch can only move data at up to 100 megabits, which is what the original numbers were showing. If you want to get closer to the direct connect speeds, you need a 1000 megabit (aka gigabit) switch instead.


Pardons for that not being exact but I edited out the duplicate gear talk. The story is that after the above testing, I bought an SMC GS8 Gigabit switch and plugged the NV+ and computer (Intel Pro 1000 switch) into the GS8 with Cat5e and re-ran as before (no jumbo frames).

write 19 MB/s
read 33 MB/s


It's not as good as some but far better than before getting the Gigabit switch. I don't understand why the direct was a little better before but am not going to sweat that. Thanks brendan and everyone.

musicfan
Message 82 of 309
lowfi
Aspirant

Re: Post your performance results

Im on wireless N right now. Copying some m4v files to my NV+

Doing around 10 or 11MB/s for writes. Reads are around 15 or so.

I use a mac pro & airport extreme N (fast ethernet). All on AFP.




I had some pictures but i cant post them because of spam prevention 😞
Message 83 of 309
bhoar
Aspirant

Re: Post your performance results

lowfi wrote:
Im on wireless N right now. Copying some m4v files to my NV+

Doing around 10 or 11MB/s for writes. Reads are around 15 or so.

I use a mac pro & airport extreme N (fast ethernet). All on AFP.


I wouldn't expect any better numbers over wireless N.

-brendan
Message 84 of 309
yoh-dah
Guide

Re: Post your performance results

bhoar wrote:
lowfi wrote:
Im on wireless N right now. Copying some m4v files to my NV+

Doing around 10 or 11MB/s for writes. Reads are around 15 or so.

I use a mac pro & airport extreme N (fast ethernet). All on AFP.


I wouldn't expect any better numbers over wireless N.

-brendan

Agree. I think these numbers are quite good for wireless.
Message 85 of 309
lowfi
Aspirant

Re: Post your performance results

I wasnt expecting more 🙂

I also plugged it directly in my mac pro and was getting around 23MB/s for writes and 45MB/s for reads. Think ill pick up a gbit switch or something and use the wireless solely for notebooks/appletv.
Message 86 of 309
yoh-dah
Guide

Re: Post your performance results

lowfi wrote:
I wasnt expecting more 🙂

I also plugged it directly in my mac pro and was getting around 23MB/s for writes and 45MB/s for reads. Think ill pick up a gbit switch or something and use the wireless solely for notebooks/appletv.

Maybe you can share your network recipe with other Mac users 😎
Message 87 of 309
dminches1
Aspirant

Can't Get Iometer to run

I followed all the instructions but I am get an error "Worker 1" on "JJG" cannot set specification NV (the name of my test file). The largest request or reply size (262144 bytes) may be too large.

Any idea what I am doing wrong?
Message 88 of 309
dminches1
Aspirant

Re: Post your performance results

Ok. I got it working but the results are not pretty.

System:
Pentium 4 (3.2 GHz); 3 GB Ram; XP Pro
Intel PRO/1000 CT (jumbo frames disabled; link speed - auto detect)
Netgear GS608 switch
ReadyNas NV running RAIDiator 4.00b9-p1-T1 [1.00a037] in XRAID; 1 GB RAM
4x Samsung 500GB ; 75% filled
Disk write cache: ENABLED
Journaling: ALL DISABLED
Fast CIFS Writes: ON


I am getting around 10 MB/s on both reads and writes. This can't be the fastest speed I can get.

What should I change to improve the speed? I was wondering if my NIC card is the bottleneck. I can't see the switch being an issue since it is a gigabit switch. I am assuming I don't need to set anything on it since the NV reports 1000Mbit / full duplex.

Can someone give me a couple suggestions?
Message 89 of 309
phrozen
Aspirant

Re: Post your performance results

I ran the bench from:
http://www.infrant.com/download/iometer.icf

System:
Dell Poweredge 400sc
Pentium 4 (2.8 GHz); 1.5 GB Ram; XP Pro
Intel PRO/1000 MT (jumbo frames enabled; link speed - auto detect; all offload options enabled)
Netgear GS105 switch
ReadyNas NV running RAIDiator™ v3.01c1-p6 in XRAID; 256 MB RAM
4x Seagate ST3750640AS 698 GB ; 93% filled
Disk write cache: ENABLED
Journaling: ALL DISABLED
Fast CIFS Writes: ON

1 Seq Read 256K: 16.411689 MBps
1 Seq Write 256K: 8.280536 MBps


Boo...

:cry:
Message 90 of 309
PhiWi1
Aspirant

Re: Post your performance results

Hardware:

Some Core2Duo Notebook by LG with Broadcom Gig-E NIC
Netgear WNR854T DSL/WLAN/GIG-E Router
ReadyNas NV+

Jumbo Frames: ON
Virus Scanner: ON

Result:
Read: 30.40MB/s
Message 91 of 309
OGSI
Aspirant

ReadyNAS 1000S

ReadyNAS 1000S
512 MB memory
4 x Seagate ST3250823AS 250 GB drives [232]
RAID Level 5, 4 disks
Jumbo frames disabled
Journaling disabled
Fast CIFS writes enabled

HP 2.6GHz P4 PC
Windows XP Pro
1GB memory
Broadcom NetXtreme Gigabit Ethernet
Flow control and TCP offload options enabled

Cisco 2960 48port switch

MB/s Write 0.4
MB/s Read 0

doen't make sense at all!
but it is very slow and the test took a long time, much longer then writing a 1GB file to a drive
Message 92 of 309
Plukkie
Aspirant

ReadyNAS NV+ performce results

NV+ Setup:
Model: ReadyNAS NV+ [X-RAID]
Firmware: RAIDiator 4.00c1-p2-T19 [1.00a037]
Memory: 1024 MB [2.5-3-3-7]
Ch 1 : Seagate ST3500630NS [465 GB] 461 GB allocated
Ch 2 : Seagate ST3500630NS [465 GB] 461 GB allocated
Ch 3 : Seagate ST3500630NS [465 GB] 461 GB allocated

- Jumbo Frames enabled
- journalling disabled
- fast CIFS writes enabled

Network / PC Setup:
- Asus switches GigaX1108N and GigaX1105N in between
- PC Intel Core2 QUAD 6600, 2x1GB DDR2 RAM PC6400, 2x WD160GB SATA300 Striping Mode
- LAN onboard Marvell Yukon 88E8056 PCI-E
- Windows XP Enterprise Edition SP2 Inegrated

PC Settings:
- Disabled Sygate Personal Firewall
- Disabled Comodo BOclean Anti Malware
- Disabled Symantec AV
===========================================================
Test description:
For READ test copied 1023 MB VOB-file from NAS to PC.
FOR WRITE test copied 1023 MB VOB-file from PC to NAS.
Both tests done with Total Commander and with CIFS protocol (Windows file sharing).

Test results:

Jumbo Disabled on PC:
- READ 194 Mbit/s
- WRITE 102 Mbit/s


Jumbo 4088 bytes on PC:
- READ 199 Mbit/s
- WRITE 127 Mbit/s


Jumbo 9014 bytes on PC:
- READ 204 Mbit/s
- WRITE 126 Mbit/s


Did my best, this is the max I can get. In the past I also discovered that uninstalling the firewall from the PC gave more throughput compared to disabling it, surprisingly. But, because I definately want to have protection, I leave it like this.
Message 93 of 309
meschubert
Aspirant

Re: Post your performance results

ReadyNAS NV+ Setup:
Model: ReadyNAS NV+ [X-RAID]
Firmware: RAIDiator 4.00c1-p1 [1.00a037]
Memory: 1024 MB [2.5-3-3-7]
RAID Config: X-RAID (Expandable RAID), 4 disks, 6% of 1272 GB used
RAID Disks: 4 x Seagate ST3500630AS [465 GB]; 461 GB allocated
File Protocols: CIFS

Jumbo Frames: Enabled (box checked)
Write Caching: Enabled (Automatic when UPS Present)
Disable full data journaling: Enabled (box checked)
Disable journaling: Disabled (box not checked)
Optimize for OS X: Disabled (box not checked)
Enable fast CIFS writes: Disabled (box not checked)
Enabled fast USB disk writes: Enabled (box checked)
Opportunistic Locking Enabled (box checked; all shares)

Network Setup:
Netgear GS108 Switch
1’ CAT 6 run to the ReadyNAS NV+
100’ CAT 6 run to the PC used for testing
2nd GS108, 50’ to 100’ runs to two other PCs, 3 DIRECTV TiVos and other devices

Relevant extracts from a component list I made up while building the testing PC a few months ago:
ABIT IP35 Pro LGA 775 Intel P35 AXT Intel Motherboard
- Includes onboard Realtek RTL8169/8110 Family Gigabit Ethernet NIC (NDIS6.0)
- 2nd NIC Disabled
Intel Q6600 quad cpu (NOT over clocked during this testing)
Critical Ballistix Tracer 4GB 240 pin SDRAM DDR2 1066 (PC2-8500 Dual Channel)
- 2 x 2 x 1GB Matched Pairs
Windows Vista Ultimate
NOD32 AV V3 installed and enabled
Comodo FW V3 installed and enabled
Comodo BOClean Anti-Malware installed and enabled

IOMeter 256k Read/Write Setup:
Configured as recommended in Hans Solo posts except Maximum Disk Size = 8192000

IO Meter Results for PC to an identical Seagate ST3500630AS drive installed internally (SATA II)
For reference, note that NV+ runs SATA II drives at SATA I drive rates; NTFS Formatting
Reads: 67.39 MB/sec Max; 63.33 MB/sec Average
Writes: 63.77 MB/sec Max; 62.74 MB/sec Average

IO Meter Results for NV+ Share (10 read runs & 10 write runs)
Reads: 34.75 MB/sec Max; 30.81 MB/sec Average
Writes: 19.15 MB/sec Max; 17.49 MB/sec Average

IO Meter Results for PC to USB Drive Attached to NV+ (10 read runs & 10 write runs)
USB Drive is a Seagate ST3500641CB-RK (500 MB); NTFS Formatting
Reads: 8.72 MB/sec Max; 8.64 MB/sec Average
Writes: 4.46 MB/sec Max; 4.14 MB/sec Average

Backup results from NV+ Share to USB Drive Attached to NV+ (NTFS format; 5 runs)
Scheduled for AM; should be no other NV+ activity
Full backups of 17,103 files, 1211 folders, 42,649 GB Total, 2.9GB per file average
Mostly MP3 files and photos
2.24 MB/sec Max (5 hours, 17 Min); 2.06 MB/sec Average (5 hours, 45 min)
I’ll test backup times on large video files when I have a chance
Message 94 of 309
wardie
Aspirant

Re: Post your performance results

Here's a comparison between XP and Vista machines on otherwise identical network/NAS:

NV+ Setup:
Model: ReadyNAS NV+ [X-RAID]
Firmware: RAIDiator 4.00c1-p2-T45 [1.00a037]
Memory: 1024 MB [2.5-3-3-7]
Volume C: Online, X-RAID (Expandable RAID), 3 disks
Ch 1 : Seagate ST3500630NS [465 GB] 461 GB allocated
Ch 2 : Seagate ST3500630NS [465 GB] 461 GB allocated
Ch 3 : Seagate ST3500630NS [465 GB] 461 GB allocated
- Jumbo Frames disabled
- journalling disabled
- fast CIFS writes enabled
- Oplocks enabled on shares
- VistaBoost on

Network Setup:
- Netgear GS108 gigabit switch
- Cat 5e

Old PC #1 test: 🙂
- Pentium 4 @ 2.8 GHz
- 1024MB RAM
- Gigabit NIC jumbo off
- Win XP Pro
- iometer test as per yoh-dah specification
Write: 22MB/s
Read: 33MB/s

CPU utilisation during tests: 40%

New PC #2 test: 😞
- Intel quad core Q6600 @ 2.8 GHz
- 2048 MB RAM
- Gigabit NIC jumbo off
- Win Vista Home Premium SP0
- iometer test as per yoh-dah specification
Write: 11.5MB/s (9.8 without VistaBoost)
Read: 6.5MB/s (6.1 without VistaBoost)
CPU utilisation during tests: 0-1%

Thanks Microsoft - I await Vista SP1.
Message 95 of 309
Lars1
Aspirant

Re: Post your performance results

According to the Iometer template I get :

Desktop - P4 3,4Ghz, 2GB ram, Gbit
Read : 19,3MB/s * should be higher
Write : 22,7MB/s * ~satisfactory

Laptop - PM 1.8Mhz, 512MB ram, 100Mbit
Read : 40 something
Write : 35++

- Strange numbers for a 100Mbit laptop connection indeed, and stays similar even if I increase the sectors to 8M :?

** Readynas NV+, almsot factory default T45 firmware, no jumboframes, no afp, domain security mode, 4x300GB seagate AS, Xraid
Message 96 of 309
ChicagoScott
Aspirant

Re: Post your performance results

Lars wrote:
... Laptop - PM 1.8Mhz, 512MB ram, 100Mbit
Read : 40 something
Write : 35++

- Strange numbers for a 100Mbit laptop connection indeed, and stays similar even if I increase the sectors to 8M :?

I found that Iometer incorrectly calculates the speed when the CPU clock isn't running at a fixed rate. You might want to try Iometer again with the laptop set at "Maximum Performance" or some other setting where the CPU doesn't slow down to reduce heat/battery drain.

-Scott
Message 97 of 309
Lars1
Aspirant

Re: Post your performance results

Ofcourse.. I was tired and didn't remember this issue :wink:

With cpu locked at maximum I got more normal results, though I find it strange I'm consistently seeing higer write than read speeds + 11MB/s is too high on a 100Mbit connection.

Laptop - PM 1.8Mhz, 512MB ram, 100Mbit
Read : 4.56 MB/s
Write : 11.02 MB/s
Message 98 of 309
damianp
Aspirant

AFP performance on OS X 10.5.1

System Info
Xbench Version 1.3
System Version 10.5.1 (9B18)
Physical RAM 4096 MB
Model MacBookPro3,1
Disk Test 27.73
Sequential 16.24
Uncached Write 6.23 3.83 MB/sec [4K blocks]
Uncached Write 54.88 31.05 MB/sec [256K blocks]
Uncached Read 16.81 4.92 MB/sec [4K blocks]
Uncached Read 123.33 61.98 MB/sec [256K blocks]
Random 94.76
Uncached Write 36.72 3.89 MB/sec [4K blocks]
Uncached Write 95.45 30.56 MB/sec [256K blocks]
Uncached Read 665.43 4.72 MB/sec [4K blocks]
Uncached Read 333.58 61.90 MB/sec [256K blocks]
Message 99 of 309
dberger
Aspirant

NFS Perf

Expanded from my previous post here: /forum/viewtopic.php?t=15033

NAS: NV+, 1GB RAM, 3x500GB disk, X-RAID, Jumbo frames enabled

Network: Netgear GigE switch, w/ Jumbo frame support

Client: AMD X2 4200+ (dual core), running Fedora Core 6 (2.6.22.2-61.fc6), intel GigE nic, mtu=7936

exportopts: insecure,insecure_locks,rw,no_root_squash,async

mountopts: proto=tcp,rsize=8192,wsize=8192,intr,bg

From my linux box:

(write test - default 1k block size)
$ dd if=/dev/zero of=/path/to/file/on/nas count=200000
200000+0 records in
200000+0 records out
102400000 bytes (102MB) copied, 15.2132 seconds, 6.7 MB/s

testing with different block sizes (4k, 8k, 16k, 256k) yields a bit of difference - up to about 8MB/s with 4k blocks.

(read test - default 1k block size)
$ dd if=/path/to/file/on/nas of=/dev/null
200000+0 records in
200000+0 records out
102400000 bytes (102MB) copied, 7.81847 seconds, 13.1 MB/s

playing with different block sizes (bs=#) here (4k, 16k) is tricky - if you re-read the same file, it's cached and you can see upwards of 500MB/sec. If I re-create the file each time, I've seen 15MB/s best case read perf using 4k blocks.

If I ssh onto the device and perform the same tests against the local file system, I get only marginal improvement on write speed - 6.8MB/s write.

Again, results differ with block sizes - best I've found is 18MB/s write w/ 4k blocks.

Read against the local file system is significantly improved, at 31.8MB/s, or 104MB/s with 4k blocks.

When I've got a bit of time, I'm going to see about getting IOMeter working with the manager running on a windows box and dynamo (the test process) on my linux box.
Message 100 of 309
Top Contributors
Announcements