This is going to take a rather focused investigation by Rogers and Twitch engineers and a willing volunteer, all working together to get to the bottom of the problem. This would be much easier if there was another modem available for test purposes that wasn't an Intel product. That would either clear the Intel modems or place the problem directly in Intel's lap. I'll have to post a question in the DSLReports TPIA forums just to see if TPIA users on Rogers Cable system have any success with Twitch uploads.
Although I'm not abel, I'm having the same issues where the stream will be rock solid for a few min, completely go to 0kb/s then recover and go back to rock solid. I'm on the CODA4852 modem and I recently switched over from Start where I had Zero problems streaming.
This is my SNB info:
and this is my Twitch inspector info:
As you can see, it flatlines throughout the entire stream and essentially freezes the stream, but the sound goes through. It's a wired connection with no extra routers or bridges or anything of the sort.
@Tal3 what modem model were you using with Start? I'm curious at this point to know if it was a Hitron CDA3 or CGN variety or something completely different. And ..... cable or DSL/VDSL? You were on R-cable with Start, correct? If I have that correct then the difference would be the modem and the back-end support as your data should have been routed to Start's servers and then onwards from there.
Thank you. Baring any huge differences in how twitch RTMP data is treated by Rogers vs Start, that lays the blame on the modem.
@RogersShaun nope, this is an engineering issue that will take concerted action by Rogers and Twitch Engineering staff to sort out. From what I can see there's nothing that I or Tech Support can do at this point in time. Using a VPN appears to be a workaround, which points to the fact that as long as the modem doesn't know that the data stream is RTMP based, then its quite satisfied to run it with any VPN data rate restrictions found in the Puma 6 and 7 firmware. As I've indicted before, this will take a test session or two with Rogers and Twitch Engineers, with full access to the modem and network in order to trace the data from the customer's modem, thru the network and finally to the Twitch servers. Even then, it will result in a list of actions that have to be completed to ensure that RTMP streaming works on the Rogers network, not only for Twitch, but also for Youtube and others. That customer will have to be a volunteer, willing to run a test stream and provide any tech details required by the Rogers and Twitch Engineers. Just a guess, this might take a few hours via three way conference call to accomplish.
The one nagging problem here is that no one appears to have details on hand that spell out specific data rates for the various protocols that the modems might encounter. There are known issues with data rates for non-TCP/IP,ICMP or UDP traffic. So, someone in the Engineering Staff needs to take the bull by the horns and sort this out as its an issue not only with streaming RTMP, but also with VPNs and with IPSEC traffic. My personal thought is that these issues might date back to the 2007/2008 timeframe when the Puma 5 was first developed by Texas Instruments. Intel bought the Puma product line from Texas Instruments in 2010, and possibly carried on any limitations by simply porting portions of the firmware to both Puma 6 and 7 modems, both of which have problems with out of the ordinary protocols.
At the end of the day, Rogers will have to make some pronouncement that they are, or are not interested in running protocols other than run of the mill TCP/IP, ICMP and UDP. Users these days expect much more out of their services, from streaming to VPNs to IPSEC and others. I'm sure that TPIAs are more than willing to step in if Rogers prefers to stay out of the game.
@Tal3 wrote:
Although I'm not abel, I'm having the same issues where the stream will be rock solid for a few min, completely go to 0kb/s then recover and go back to rock solid. I'm on the CODA4852 modem and I recently switched over from Start where I had Zero problems streaming.
This is my SNB info:
and this is my Twitch inspector info:
As you can see, it flatlines throughout the entire stream and essentially freezes the stream, but the sound goes through. It's a wired connection with no extra routers or bridges or anything of the sort.
as @Datalink mentioned, the issue is in Rogers Routing. However, your signal levels are very weak, and the SNR numbers are low. I would suggest sending a PM to @CommunityHelps asking them to check your signal levels and SNR, then to compare to your neighbours to see if your low levels are neighbourhood wide, or only affecting your modem.
Either way they should send a tech out to improve the SNR levels.
I'm waiting on my new lan card in 2 days since when I look at most people that have this problem most of them got killer network adapters so I'm a rule out that before I go in further details and since I heard of some people changing helped I don't want to waste no ones time you know but here is my levels anyways
At this point in time I really, really don't believe that its a case of intentional throttling. All of the evidence that I've seen so far points to a protocol issue with the Puma series modems. Another poster noted that he was on Start, using a non-Puma modem and didn't have any issues, and that was with a service that used the Rogers network. So, this can work, on Rogers network, with a non-Puma modem. Its really unfortunate that Rogers doesn't offer such a modem as that could instantly indicate whether or not the problem is with the Puma series modems or not.
So, I'm not saying that the issue of a possible violation of net neutrality is inappropriate. I just don't believe that its the case here. But, I am absolutely willing to look at any evidence that indicates otherwise.
As I've indicted elsewhere, testing consumer modems is sorely lacking, personal opinion. That includes testing all of the various protocols that are legal for use, as indicated by their respective RFCs. That should be done by Intel before the modem is ever released for manufacturing. That would avoid situations such as this, or, personal opinion, it should avoid situations such as this. Should and would are two completely different concepts here.
Just to note, the days of a simple modem are long passed. These days, DOCSIS modems do a lot more than one would think.
-
45396 views
-
21 Likes
-
-
