Edgerouter PRO and Router Charts

[coxhaus]

Has there been any discussion on why the EdgeRouter PRO shows slower on the router charts than a EdgeRouter Lite?

I would like an answer before I would buy one.

 

[Cloud200]

Has there been any discussion on why the EdgeRouter PRO shows shower on the router charts than a EdgeRouter Lite?

I would like an answer before I would buy one.

It looks like the review was done by two different people using two different sets of equipment.
It is possible the second review messed up the config somehow.

I have a few ER8-Pros out in the field. I can confirm 100% that connecting via gigabit fiber they will do full duplex wirespeed NAT routing with Fastpath enabled.
Without, its about 1100 mb/s total up/down.

 

[coxhaus]

I here what you are saying but we have been tuned to use the router graphs for making decisions. This is what was published. Is there a fix coming for the router graphs? Do we now disregard router graphs because there may be errors.

I would still like to see the real speed in the router graphs.

 

[Cloud200]

I here what you are saying but we have been tuned to use the router graphs for making decisions. This is what was published. Is there a fix coming for the router graphs? Do we now disregard router graphs because there may be errors.

I would still like to see the real speed in the router graphs.

Ahh well in that case, the only ones that could do that are doug or tim . . .
Paging @thiggins
Paging @dreid

 

[System Error Message]

I have both the edgerouter pro and the mikrotik CCR1036. The CCR is louder and costs more but is a lot better than the edgerouter. if you want the best router go for mikrotik. If you want something versatile go for ubiquiti as you can run linux programs on it.

I dont really like the ubiquiti edgerouter's firmware but it i still better than consumer routers. Just avoid the ERL because of the flash issue.

The CCR is also more reliable than edgerouter because after leaving the edgerouter as it is i cant even access it now. network settings are all correct.

 

[thiggins]

Data posted in all the Charts is based on tests done by me, using the stated test procedures. That's the only way that proper comparative results can be presented.

The Lite and Pro were tested with the same process. If you look at the IxChariot plots in the review, you see high throughput variation on downlink. That's the reason for the low downlink result. What the cause of that is/was, I don't know. No retest is planned and nothing is wrong with the Router Charts themselves.

The review also states that both the Lite and Pro are capable of throughput beyond our ability to accurately test, i.e.

The throughput results listed in Table 2 are obviously not reflective of the router's actual 8 Gbps spec'd performance.

So, please read the product review IN ADDITION to using the Charts. As for your buy decision, either product has way more performance than you'll probably be able to use.

 

[thiggins]

Ahh well in that case, the only ones that could do that are doug or tim . . .
Paging @thiggins
Paging @dreid

If you need my attention, please start a conversation. I don't read every thread in the Forums. Thanks.

 

[System Error Message]

Data posted in all the Charts is based on tests done by me, using the stated test procedures. That's the only way that proper comparative results can be presented.

The Lite and Pro were tested with the same process. If you look at the IxChariot plots in the review, you see high throughput variation on downlink. That's the reason for the low downlink result. What the cause of that is/was, I don't know. No retest is planned and nothing is wrong with the Router Charts themselves.

The review also states that both the Lite and Pro are capable of throughput beyond our ability to accurately test, i.e.


So, please read the product review IN ADDITION to using the Charts. As for your buy decision, either product has way more performance than you'll probably be able to use.

I am planning to test the edgerouter PRO and i will do my best to test these high performance routers to the limit so the delay is just getting the test bed set up and having to create a hacking program to flood the routers.

 

[thiggins]

I am planning to test the edgerouter PRO and i will do my best to test these high performance routers to the limit so the delay is just getting the test bed set up and having to create a hacking program to flood the routers.

Wonderful. Please share the testbed and test process details when you publish results.

 

[Cloud200]

If you need my attention, please start a conversation. I don't read every thread in the Forums. Thanks.

Ahhhh, my mistake.
I was under the impression that was what the "you were mentioned in a post" system was for.

Thanks

 

[coxhaus]

I did read the review you posted. My impression from reading the review is the high through put potential is related to the multiple port back plane kind of like a switch but you only test a couple of ports.. Maybe because of all the ports and you are only testing a couple of ports, your software is picking up a time slice which shows as a slower speed.

The other thing I found interesting is you state the following "The review also states that both the Lite and Pro are capable of throughput beyond our ability to accurately test " What is your thinking? Can you explain? As it makes me wonder whether you think there is more parallel processing in the box or there is maybe more code tuning to be done. I am left wondering.

The Edgerouter lite and pro compared based on speed without hardware acceleration from what I can gather on from others on this site is the pro is not twice as fast as the lite maybe 1/2 again as much. I don't want to compare fastpath as I am sure I would do something to cause the router to drop out of fastpath. I like playing with routers.

PS
The only reason I brought this up is at the end of the review there is a "discuss this in the forums".

I did do a search under routers and did not find any discussion. It turns out the discussion is under LAN & WAN. All I had to do is click the link.

 

[thiggins]

If you look at the EdgeRouter spec sheet, the high packet per second specs on the product page are for very small 64 Byte packets. For 512 Byte and larger packets, they spec the total of Gigabit line rate for all eight ports.

Our current test method using IxChariot and Windows 7 can't accurately measure unidirectional throughput past around 800 Mbps. This reflects limitations of the TP-LINK TG-3468 NICs, Windows, internal computer bus structure, drivers, etc.

So when I see throughput that high, I usually state in the review that the product routes at Gigabit wire-speed.

Testing Gigabit wire speed on a single port, let alone 8 simultaneously, requires specialized test gear beyond our budget.

If you seek more detailed answers, try the Ubiquiti forums.

 

[System Error Message]

Assuming that the firmware is good i see the edgerouter pro being twice as fast as the ERL because the CPU is twice as fast in frequency even though it is the same CPU. I already have more than 8 NICs available to test the ports but i am not going to be testing layer 3 routing, rather i will be testing NAT performance and connection limits.

To 99% of users here NAT performance matters more than layer 3 routing.

 

[coxhaus]

Seems to me there must be a problem with their firmware where 2 ports can not fully utilize the processor otherwise the router graph charts would be different. I am sure it will be faster if you have lots of WAN connections which is probably what it was built for. I was just looking for a faster processor in a router to use with one WAN port. I want a faster processor to make up for when I use the router without hardware acceleration. I guess I am looking for a router which will process a gig connection without hardware acceleration.