What is being done to address the high latency issue with Hitron Modems and is there any forecast for a date when a fix will be seen in a firmware update? This has also been noticed on the Shaw versions of the modem.
@Datalink. Great question, the issue is actually as a result of the increased number of channel that we are bonding and is not specific to the Hitron device. Having said that we did publish our latency results by product in the latest SamKnows report (www.samknows.com/rogers ) relative to our American peers and we are consistently offering a better latency result. If there are specific sites where you are seeing increased ping times, we would love to hear about these to see what we can do to optimize them, please post on the internet boards.
I was questioning the latency with the Hitron Modems themselves. If you do nothing but ping the modem, with essentially no internet traffic, the returned ping times are unstable and can have a very wide range. The Sam-Knows data is interesting to say the least, with ping times in the 30 ms range. The modem itself can have a ping time of 20 ms or more. Firmware version 18.104.22.168 for the CGN3 would exhibit ping times of up to 100 ms.
In a nutshell, there are two issues here:
1. The latency to and thru the modem, which to some degree will be tied to how a TDMA system works; and
2. Latency to sites such as League of Legends, World of Warcraft and others, which to some degree depends on traffic to those sites, how well those sites handle the traffic and peering agreements with those sites. The gamers in the crowd would be interested in peering agreements, latency issues to those sites, and what can or will be done to alleviate the latency to those sites. Without knowing what sites the Sam-Knows box uses, I suspect that if you plugged in few of the gaming sites for test purposes, you would see results which might be considerably worse than what is reported.
@Datalink thanks for the great detail in your response; I will definitely discuss with SamKnows what they are using to measure their latency times. As for the latency within the modem we are currently on 22.214.171.124 which should have addressed a number of the issues, I will however follow up with Hitron to have them confirm that the latency within the modem has decreased.
If you ping a CGN3 a thousand times, the average latency will be in the 4 to 5 ms range, with a rare occasional minimum of 1 ms and the occasional max in the 20 ms range. If you ping a router, you will see 1 ms or less typically. Now, the modem is doing a lot more that running as a router, I'll say that much, but, the times that are observed lead one to think, ok, there is a firmware issue still afoot, or the modems are underpowered for what they are asked to do.
There is an interesting comment on the DSL Reports Shaw forum which states:
"The Hitron deprioritizes ICMP traffic causing any ICMP echo (ping) testing to be highly inaccurate.
Ookla speed tests (speedtest.net, speed.shaw.ca) use TCP based latency testing. You will noticed those tests show an accurately low latency.
So the ICMP numbers you see are artificially high, the actual latency for TCP data flow is typical 8ms - 12ms for the first Shaw hops. Unfortunately so many things (network performance testing, online gaming) depend on the ICMP latency figures, so even though the TCP latency is nice and low, the poor handling of ICMP (and maybe UDP) traffic needs to be solved.
Shaw engineering is aware and the vendor is working for a firmware resolution."
Where this info comes from I don't know, but it is interesting. I for one would be really interested in whether or not there is any truth to it and whether or not Hitron has this situation in hand. Is what we are seeing in modem latency the best it will ever be, or is there really a solution to the problem?
@Datalinkas mentioned in one of my earlier responses to you Hitron brings fixes from other ISPs and Intel as well as specific requests from Rogers. I will definitely follow up with Hitron to get their understanding of this issue and if there is a fix we will have it added into a future firmware release.