Posted: Thu Jan 20, 2011 4:14 Post subject: Cisco Linksys E2000 - routing performance problem
Description:
I have performance problems, which seem to be related to the routing engine and/or my configuration. I would appreciate any advise.
Configuration:
- Physical ports #1-3 (logical #2-4) are in assigned to vlan1 (LAN, vlan1ports="2 3 4 8*")
- Physical port #4 (logical #1) is assigned to vlan3 (vlan3ports="1 8")
- WAN port is assigned to vlan2 (vlan2ports="0 8")
- QoS and UPnP are disabled
- br0 interface assigned with 192.168.0.1 IP address
- vlan3 interface assigned with 192.168.1.1 IP address
Test scenario #1:
PC is connected to port #1 link (here and below I refer to physical ports), Samba server is connected to port #2, both have IP addresses from the same subnet (192.168.0.x) and 1Gbit links. E2000 acts as a simple network switch.
When I am transferring files from the server to the PC, transfer speed is between 12MB/s and 16MB/s (megabytes per second). Not perfect, but good enough.
Test scenario #2:
Moved Samba server from port #2 to port #4, reassigned it's IP address to 192.168.1.2. E2000 now acts as a router between 2 subnets. PC is still connected to physical port #1.
Transfer speed dropped down to 4-5MB/s (3 times !).
Versions:
I have tried 2 versions (followed by 30-30-30 resets) and got similar results in both cases:
- dd-wrt.v24-14929_NEWD-2_K2.6_std_usb_ftp-e2000.bin
- dd-wrt.v24-15943_NEWD-2_K2.6_std_usb_nas-e2000.bin
I am also attaching a file, that contains vlan part of my NVRAM configuration, if you want to take a look.
That is about the limit of how fast the E2000 can route traffic. Switching is fast and handled by dedicated ASIC's, routing is slow and handled by software. _________________ Read the forum announcements thoroughly! Be cautious if you're inexperienced.
Available for paid consulting. (Don't PM about complicated setups otherwise)
Looking for bricks and spare routers to expand my collection. (not interested in G spec models)
That is about the limit of how fast the E2000 can route traffic. Switching is fast and handled by dedicated ASIC's, routing is slow and handled by software.
Thanks for the answer, this is sad, though. Is this a CPU or internal bus speed limitation ? I wouldn't upgrade my WRT160N to E2000, if I knew that.
Joined: 24 Aug 2009 Posts: 2070 Location: South Florida
Posted: Thu Jan 20, 2011 21:10 Post subject:
VladStar wrote:
phuzi0n wrote:
That is about the limit of how fast the E2000 can route traffic. Switching is fast and handled by dedicated ASIC's, routing is slow and handled by software.
Thanks for the answer, this is sad, though. Is this a CPU or internal bus speed limitation ? I wouldn't upgrade my WRT160N to E2000, if I knew that.
This has proven to be fantastic on both my WRT320N and E2000:
http://tomatousb.org/download _________________ Optware, the Right Way
Asus RT-AC68U
Asus RT-N66U
Asus RT-N10
Asus RT-N12
Asus RT-N16 x5
Asus WL520gU
Engenious ECB350
Linksys WRT600Nv1.1
Linksys WRT610Nv1
Linksys E2000
Netgear WNDR3300
SonicWall NSA220W
SonicWall TZ215W
SonicWall TZ205W
SonicWall TZ105W
TomatoUSB doesn't seem to support VLAN's. The whole idea was to have a server in a separate VLAN, so nobody can break into my LAN PC computers, in case if server gets compromised.
TomatoUSB doesn't seem to support VLAN's. The whole idea was to have a server in a separate VLAN, so nobody can break into my LAN PC computers, in case if server gets compromised.
It does, it's just not implemented into the GUI..yet
It also supports Jumboframes.. _________________ Optware, the Right Way
Asus RT-AC68U
Asus RT-N66U
Asus RT-N10
Asus RT-N12
Asus RT-N16 x5
Asus WL520gU
Engenious ECB350
Linksys WRT600Nv1.1
Linksys WRT610Nv1
Linksys E2000
Netgear WNDR3300
SonicWall NSA220W
SonicWall TZ215W
SonicWall TZ205W
SonicWall TZ105W
FYI even though Tomato has a bit better routing throughput, routing will still be far slower than switching. _________________ Read the forum announcements thoroughly! Be cautious if you're inexperienced.
Available for paid consulting. (Don't PM about complicated setups otherwise)
Looking for bricks and spare routers to expand my collection. (not interested in G spec models)
I've upgraded from G300N to E2000.
I get around 8.8MB/s routing throughput (with dd-wrt),
which is a bit pathetic compared to 100Mbit
G300N where I was getting close to 11MB/s (also with dd-wrt).
The CPU usage is max 30%. Unfortunately I did
not test with original firmware. How
much better would the tomato be on E2000 ?
FYI even though Tomato has a bit better routing throughput, routing will still be far slower than switching.
Flashed TomatoUSB (version 1.28 ), configured with the same set of settings as described at the top of this topic and tested within the same environment - got 7-7.5MB/s transfer speed, which is 1.5 times faster than DD-WRT performed.
Looks like there is some overhead in a code, that makes DD-WRT slower, compared to TomatoUSB.
Last edited by VladStar on Mon Jan 24, 2011 1:50; edited 1 time in total
I've upgraded from G300N to E2000.
I get around 8.8MB/s routing throughput (with dd-wrt),
which is a bit pathetic compared to 100Mbit
G300N where I was getting close to 11MB/s (also with dd-wrt).
The CPU usage is max 30%. Unfortunately I did
not test with original firmware. How
much better would the tomato be on E2000 ?
You didn't say an exact model for whatever "G300N" is, but many Ralink models and some Atheros models have switches that are capable of layer 3 switching (or something very similar to it) which makes them have practically no CPU load from routing.
In regard to the 30% CPU usage claim please see this.
Flashed TomatoUSB (version 1.28 ), configured with the same set of settings as described at the top of this topic and tested within the same environment - got 7-7.5MB/s transfer speed, which is 150% times faster than DD-WRT performed.
Looks like there is some overhead in a code, that makes DD-WRT slower, compared to TomatoUSB.
Your wording is poor/inaccurate. Based on your numbers it is ~50% faster now, or in other words 1.5 times faster, but "150% times faster" is gibberish. _________________ Read the forum announcements thoroughly! Be cautious if you're inexperienced.
Available for paid consulting. (Don't PM about complicated setups otherwise)
Looking for bricks and spare routers to expand my collection. (not interested in G spec models)
Your wording is poor/inaccurate. Based on your numbers it is ~50% faster now, or in other words 1.5 times faster, but "150% times faster" is gibberish.
Sorry, meant to say "1.5 times faster" or "50% faster", but ended up with a mix.
it was buffalo WHR-G300N , I was unable
to configure separate lan subnets on it hence
the upgrade to E2000. E2000 consumes double
the power and is slower
anyway I tried tomato, it's a bit faster
(about 10% in my case) but when there is a heavy
traffic between lans (vlan1 and vlan3), the lan to wan througput drops to zero in tomato. No such
problem in dd-wrt.
ive tested the E2000 with DD-WRT and between WAN - LAN i run iperf
about 88Mbps troughput, before with the orginal stock firmware it was about 199Mbps trougput.
can somebody solve this issue?
Only if you want to throw away many of the features of custom firmwares. Stock firmware uses a different method of NAT that is simpler and faster but because of its simplicity it isn't capable of NAT'ing certain kinds of traffic correctly. The stock firmware doesn't support any of those kinds of traffic so it is free to use this alternative method. _________________ Read the forum announcements thoroughly! Be cautious if you're inexperienced.
Available for paid consulting. (Don't PM about complicated setups otherwise)
Looking for bricks and spare routers to expand my collection. (not interested in G spec models)