SHO VS Real Router

[Mir96TA]

I was playing with my SHO Router with Multilevel Service Router (EOL) from Cisco.
I see a world class difference.
Since Cisco E4200 v1 is a SHO router it can spoof you its doing good; until you put a load on it.

Check this out
First off all base
Base Line Maximum Hardware limitations.
Lan to Lan GigE Switch connection

D:\iperf>iperf -c 192.168.1.110 -f m -P 1 -i 2 -t 20 -w 1M
------------------------------------------------------------
Client connecting to 192.168.1.110, TCP port 5001
TCP window size: 1.00 MByte
------------------------------------------------------------
[ 3] local 192.168.1.100 port 49438 connected with 192.168.1.110 port 5001
[ ID] Interval Transfer Bandwidth
[ 3] 0.0- 2.0 sec 225 MBytes 945 Mbits/sec
[ 3] 2.0- 4.0 sec 226 MBytes 947 Mbits/sec
[ 3] 4.0- 6.0 sec 226 MBytes 949 Mbits/sec
[ 3] 6.0- 8.0 sec 226 MBytes 948 Mbits/sec
[ 3] 8.0-10.0 sec 226 MBytes 948 Mbits/sec
[ 3] 10.0-12.0 sec 226 MBytes 948 Mbits/sec
[ 3] 12.0-14.0 sec 228 MBytes 956 Mbits/sec
[ 3] 14.0-16.0 sec 226 MBytes 948 Mbits/sec
[ 3] 16.0-18.0 sec 226 MBytes 948 Mbits/sec
[ 3] 18.0-20.0 sec 226 MBytes 948 Mbits/sec
[ 3] 0.0-20.0 sec 2262 MBytes 948 Mbits/sec

====================================
Following test are based on NAT/PAT performance.
Router Cisco E4200 iOS DD-WRT v24-sp2 (12/05/13) mega - build 22786

Lan => Lan GigE N UDP Nat Test


D:\iperf>iperf -u -c 192.168.1.110 -f m -P 1 -i 2 -t 20 -w 1M -b 1000M
------------------------------------------------------------
Client connecting to 192.168.1.110, UDP port 5001
Sending 1470 byte datagrams
UDP buffer size: 1.00 MByte
------------------------------------------------------------
[ 3] local 192.168.1.106 port 51125 connected with 192.168.1.110 port 5001
[ ID] Interval Transfer Bandwidth
[ 3] 0.0- 2.0 sec 195 MBytes 816 Mbits/sec
[ 3] 2.0- 4.0 sec 196 MBytes 821 Mbits/sec
[ 3] 4.0- 6.0 sec 196 MBytes 820 Mbits/sec
[ 3] 6.0- 8.0 sec 196 MBytes 821 Mbits/sec
[ 3] 8.0-10.0 sec 196 MBytes 820 Mbits/sec
[ 3] 10.0-12.0 sec 196 MBytes 820 Mbits/sec
[ 3] 12.0-14.0 sec 196 MBytes 821 Mbits/sec
[ 3] 14.0-16.0 sec 196 MBytes 820 Mbits/sec
[ 3] 16.0-18.0 sec 196 MBytes 821 Mbits/sec
[ 3] 18.0-20.0 sec 196 MBytes 820 Mbits/sec
[ 3] 0.0-20.0 sec 1955 MBytes 820 Mbits/sec
[ 3] Sent 1394469 datagrams
[ 3] Server Report:
[ 3] 0.0-20.2 sec 1254 MBytes 520 Mbits/sec 14.929 ms 499729/1394445 (36%)
[ 3] 0.0-20.2 sec 1 datagrams received out-of-order




D:\iperf>iperf -u -c 209.141.205.135 -f m -P 1 -i 2 -t 20 -w 1M -b 1000M
------------------------------------------------------------
Client connecting to 209.141.205.135, UDP port 5001
Sending 1470 byte datagrams
UDP buffer size: 1.00 MByte
------------------------------------------------------------
[ 3] local 192.168.1.100 port 52267 connected with 209.141.205.135 port 50
[ ID] Interval Transfer Bandwidth
[ 3] 0.0- 2.0 sec 195 MBytes 817 Mbits/sec
[ 3] 2.0- 4.0 sec 196 MBytes 821 Mbits/sec
[ 3] 4.0- 6.0 sec 196 MBytes 821 Mbits/sec
[ 3] 6.0- 8.0 sec 196 MBytes 821 Mbits/sec
[ 3] 8.0-10.0 sec 196 MBytes 821 Mbits/sec
[ 3] 10.0-12.0 sec 196 MBytes 821 Mbits/sec
[ 3] 12.0-14.0 sec 196 MBytes 821 Mbits/sec
[ 3] 14.0-16.0 sec 196 MBytes 821 Mbits/sec
[ 3] 16.0-18.0 sec 196 MBytes 821 Mbits/sec
[ 3] 18.0-20.0 sec 196 MBytes 821 Mbits/sec
[ 3] 0.0-20.0 sec 1956 MBytes 820 Mbits/sec
[ 3] Sent 1395306 datagrams
[ 3] Server Report:
[ 3] 0.0-20.2 sec 462 MBytes 191 Mbits/sec 15.461 ms 1065782/1395304
)

Here comes the TCP connection (And it took the Dive)


D:\iperf> Lan to WAN TCP connection. 1 GiG connection.



D:\iperf>iperf -c209.141.205.135 -f m -P 1 -i 2 -t 20 -w 1M
------------------------------------------------------------
Client connecting to 209.141.205.135, TCP port 5001
TCP window size: 1.00 MByte
------------------------------------------------------------
[ 3] local 192.168.1.100 port 62734 connected with 209.141.205.135 port 5001
[ ID] Interval Transfer Bandwidth
[ 3] 0.0- 2.0 sec 24.1 MBytes 101 Mbits/sec
[ 3] 2.0- 4.0 sec 27.4 MBytes 115 Mbits/sec
[ 3] 4.0- 6.0 sec 27.2 MBytes 114 Mbits/sec
[ 3] 6.0- 8.0 sec 27.4 MBytes 115 Mbits/sec
[ 3] 8.0-10.0 sec 24.4 MBytes 102 Mbits/sec
[ 3] 10.0-12.0 sec 27.4 MBytes 115 Mbits/sec
[ 3] 12.0-14.0 sec 27.5 MBytes 115 Mbits/sec
[ 3] 14.0-16.0 sec 27.2 MBytes 114 Mbits/sec
[ 3] 16.0-18.0 sec 27.1 MBytes 114 Mbits/sec
[ 3] 18.0-20.0 sec 27.4 MBytes 115 Mbits/sec
[ 3] 0.0-20.0 sec 267 MBytes 112 Mbits/sec

=============================
Here comes the Good ol Faith Cisco 3745 EOL Router
===============================
3745 Cisco Multiservice Router
Network is Lan to Wan NAT

D:\iperf>iperf -u -c 209.141.202.231 -f m -P 1 -i 2 -t 20 -w 1M -b 1000M
------------------------------------------------------------
Client connecting to 209.141.202.231, UDP port 5001
Sending 1470 byte datagrams
UDP buffer size: 1.00 MByte
------------------------------------------------------------
[ 3] local 192.168.202.253 port 53279 connected with 209.141.202.231 port 5001
[ ID] Interval Transfer Bandwidth
[ 3] 0.0- 2.0 sec 22.3 MBytes 93.5 Mbits/sec
[ 3] 2.0- 4.0 sec 22.7 MBytes 95.2 Mbits/sec
[ 3] 4.0- 6.0 sec 22.7 MBytes 95.2 Mbits/sec
[ 3] 6.0- 8.0 sec 22.7 MBytes 95.2 Mbits/sec
[ 3] 8.0-10.0 sec 22.7 MBytes 95.2 Mbits/sec
[ 3] 10.0-12.0 sec 22.7 MBytes 95.2 Mbits/sec
[ 3] 12.0-14.0 sec 22.9 MBytes 95.9 Mbits/sec
[ 3] 14.0-16.0 sec 22.7 MBytes 95.2 Mbits/sec
[ 3] 16.0-18.0 sec 22.7 MBytes 95.3 Mbits/sec
[ 3] 18.0-20.0 sec 22.6 MBytes 94.9 Mbits/sec
[ 3] 0.0-20.0 sec 227 MBytes 95.1 Mbits/sec
[ 3] Sent 161725 datagrams
[ 3] Server Report:
[ 3] 0.0-20.1 sec 227 MBytes 94.7 Mbits/sec 0.955 ms 0/161724 (0%)
[ 3] 0.0-20.1 sec 1 datagrams received out-of-order

D:\iperf>iperf -c 209.141.202.231 -f m -P 1 -i 2 -t 20 -w 1M
------------------------------------------------------------
Client connecting to 209.141.202.231, TCP port 5001
TCP window size: 1.00 MByte
------------------------------------------------------------
[ 3] local 192.168.202.253 port 49255 connected with 209.141.202.231 port 5001
[ ID] Interval Transfer Bandwidth
[ 3] 0.0- 2.0 sec 22.6 MBytes 94.9 Mbits/sec
[ 3] 2.0- 4.0 sec 22.6 MBytes 94.9 Mbits/sec
[ 3] 4.0- 6.0 sec 21.9 MBytes 91.8 Mbits/sec
[ 3] 6.0- 8.0 sec 22.1 MBytes 92.8 Mbits/sec
[ 3] 8.0-10.0 sec 22.5 MBytes 94.4 Mbits/sec
[ 3] 10.0-12.0 sec 22.5 MBytes 94.4 Mbits/sec
[ 3] 12.0-14.0 sec 22.6 MBytes 94.9 Mbits/sec
[ 3] 14.0-16.0 sec 22.0 MBytes 92.3 Mbits/sec
[ 3] 16.0-18.0 sec 22.2 MBytes 93.3 Mbits/sec
[ 3] 18.0-20.0 sec 22.4 MBytes 93.8 Mbits/sec
[ 3] 0.0-20.0 sec 224 MBytes 93.7 Mbits/sec

Keep in mind on 3745 Nic are FE not a GigE.
But here is the kicker there is not a difference between UDP and TCP traffic if you minus the overhead of TCP.

Now I know why E4200 take a Dump when ever if I do any Torrents !

 

[imagoon]

Not sure what you are trying to show here. E4200 $179.00 compared to an EOL router that was 5-15k depending on interface boards? I would think the EOL one would win hands down especially a 3745.

 

[Mir96TA]

Well what I was even an outdated Multiservices router out perform today's expensive sho router.

 

[imagoon]

Still doesn't surprise me really. Lot of those routers have more RAM, CPU and Flash than the new SHO routers. They are also purpose built rather than some small underpowered CPU running a Linux derivative and IPtables.

 

[Fardringle]

A 40-year-old Ford Mustang will still outrun a 2014 Ford Fiesta. And an old rusted out big rig will tow more than the Fiesta, too. New doesn't always mean bigger/better/faster, but small and cheap sometimes does have a purpose (i.e. basic home use).

 

[VirtualLarry]

It's "SOHO", not "SHO".

 

[Mir96TA]

It's "SOHO", not "SHO".

byte me
I know I was shorting little bit more

A 40-year-old Ford Mustang will still outrun a 2014 Ford Fiesta. And an old rusted out big rig will tow more than the Fiesta, too. New doesn't always mean bigger/better/faster, but small and cheap sometimes does have a purpose (i.e. basic home use).

I dunno about Fiesta comparesion is accurate or not.
E4200 does have 480 MHz proc
3745 is not Carrier grade router and its run @ 325Mhz
However I am impress with it

 

[alkemyst]

It's more bandwidth potential than CPU power.

I had the e4200 v1 and it's a solid SOHO device.

I have a Cisco C819 now, but that is a $2000 device.

 

[azazel1024]

Just depends on what you are doing. Does it have NAT offload, what kind of WAN connection, etc.

I've seen $80 routers that could barely handle 100Mbps PPPoE, yet does 800Mbps WAN to LAN for basic ethernet WAN connection types. At the same time, I've seen $50 routers with good offload that did 700Mbps ethernet WAN to LAN and did 680Mbps with PPPoE because they had good hardware accelerated NAT.

I've also seen $300 SMB routers that could barely push 200Mbps WAN to LAN, but could also do 80Mbps VPN.

 

[Mir96TA]

It's more bandwidth potential than CPU power.

I had the e4200 v1 and it's a solid SOHO device.

I have a Cisco C819 now, but that is a $2000 device.

My E4200 often need restart/Reset specially if I run torrent for a night.
Usally 5 Ghz goes down first then 2.4 then DHCP functions

Just depends on what you are doing. Does it have NAT offload, what kind of WAN connection, etc.

I've seen $80 routers that could barely handle 100Mbps PPPoE, yet does 800Mbps WAN to LAN for basic ethernet WAN connection types. At the same time, I've seen $50 routers with good offload that did 700Mbps ethernet WAN to LAN and did 680Mbps with PPPoE because they had good hardware accelerated NAT.

I've also seen $300 SMB routers that could barely push 200Mbps WAN to LAN, but could also do 80Mbps VPN.

I have no idea E4200 it has hardware accelerated NAT. (BTW what is hardware accelerated NAT.) I always understand NAT is PURE software function!

 

[azazel1024]

Software is math and math can be hardware accelerated. Of course really advanced mathematical functions are rather hard to accelerate.

Basic NAT can be accelerated no problem. Its translating in incoming and outgoing connection based on set parameters that the router knows. I can't say I could design such a hardware function, but it is relatively trivial to do.

Downside is, hardware accelerated NAT, even if implemented properly (which isn't always the case), generally can only be done on a basic level. Things like port forwarding, QoS, etc generally break hardware NAT. Pretty much anything that is going to involve direct OS involvement in the handling of packets will break hardware NAT.

 

[Mir96TA]

for me hardware NAT mean it is working at WIRE speed or Real Time or on the FLY. Usually some thing like ASIC.
Usually for those as you have said they do very simple things.
NAT/PAT is any thing but simple.
I never have a experience with hardware NAT/PAT so I can't say how well that works.
I am sure for VoIP or torment crowd will love it cause their packets are small but lot of them.

 

[azazel1024]

for me hardware NAT mean it is working at WIRE speed or Real Time or on the FLY. Usually some thing like ASIC.
Usually for those as you have said they do very simple things.
NAT/PAT is any thing but simple.
I never have a experience with hardware NAT/PAT so I can't say how well that works.
I am sure for VoIP or torment crowd will love it cause their packets are small but lot of them.

A lot of routers have hardware NAT. Some of the implementations work well (Qualcomm/Atheros implementations seem to work well, especially with PPPoE) others do not (Broadcom implementations don't seem to handle PPPoE).

Actually basic NAT is rather basic. It doesn't mean it doesn't take some processing power behind it once you are talking a billion bits per second and hundreds of thousands of packets per second, but the process of doing network address translation is rather straight forward.

It is when you get in to things like port forwarding, DPI and so forth that things get complex and likely/probably/can't be hardware accelerated.