Reduction of Internet latency in sessions... - Printable Version +- JamKazam Forums (https://forum.jamkazam.com) +-- Forum: Jamkazam Forums (https://forum.jamkazam.com/forumdisplay.php?fid=1) +--- Forum: New Feature Requests (https://forum.jamkazam.com/forumdisplay.php?fid=11) +--- Thread: Reduction of Internet latency in sessions... (/showthread.php?tid=964) Pages:
1
2
|
Reduction of Internet latency in sessions... - StuartR - 05-23-2020 I know that the JK team recently announced they were forging ahead on their project to somehow reduce the biggest issue for JK users... Internet latency. Anyone hazard a technical guess about how this could be accomplished? They're already using UDP peer to peer connections. Is there another protocol that could be used that could be run on a lower level (level 3?) of the IP network? The top contributors to delay in packet networks are propagation delay (i.e., the speed of light in fiber), queuing delay, and serialization delay (the time to write packets to the wire). Seems like only the last two of these are candidates for improvement. Just trying to start a discussion in advance of the JK team announcing something specific. RE: Reduction of Internet latency in sessions... - Hans Peter Augustesen - 05-23-2020 It "sounds" very interesting with that lowering latency project. Whether it is about 3 ms or 15 Here you can read about the currrent technic: Packet Rate Control And Related Systems For Interactive Music Systems > http://www.freepatentsonline.com/y2015/0256473.html As you can see, the inventors are the people from JamKazam. Yes, the "propagation delay" is very important. The average speed of the transmission on the internet is not full lightspeed, so to speak. It's more half speed. I dont think there is much to do about that. As you also mention. RE: Reduction of Internet latency in sessions... - GDJ - 05-25-2020 Agree with original posters comments. I am not expecting them to pull a rabbit out of the hat. I have contributed to the gofundme effort for a couple of reasons even though they say in so many words, they have something to try https://www.gofundme.com/f/cy2fsq-keep-jamkazam-alive-amp-improving I have used the app since March and met some folks who have been using it for years. Sounds like the Devs at one point went back to their day jobs years ago and kept the lights on, and now the renewed interest and deman. My contribution is twofold for their efforts and my usage. I take it on good faith they will make an effort to make things more stable. What to do. Unless they can run a proprietary infrastructure and wire the planet with something like fiber-optic for music only, we’re are still left to sharing the same cables with everything else. And, I’m not sure if they own a satellite (lol). In my case, as social isolation lifts it will ease the immediate concern since my playing partners are local. But in the long term, what I think needs to happen is; contracts need to be made with the ISPs to setup separate networks to customers specifically for musicians. This is the infrastructure part. In fact, I spoke with my ISP and ask if they had any sub-network we can pay for on a monthly basis so my music buddy and I could use it for playing. Think of it, no other internet traffic to content with. RE: Reduction of Internet latency in sessions... - StuartR - 05-25-2020 (05-25-2020, 03:56 PM)Grateful Dead Jams Wrote: Agree with original posters comments. I am not expecting them to pull a rabbit out of the hat. I have contributed to the gofundme effort for a couple of reasons even though they say in so many words, they have something to try RE: Reduction of Internet latency in sessions... - StuartR - 05-26-2020 Here's another random thought... Since none of us have any control over the Internet portion of total latency, this leaves the focus on the p2p client connections. What if the number of connections between each client in a session were doubled from one to two. You would send the same audio packets on each connection and the first packet that arrives at each of the other clients is used and the other one discarded. This along with dynamic packet size and frame rate adjustments might make a difference. How much would have to be tested and measured. RE: Reduction of Internet latency in sessions... - StuartR - 05-31-2020 Well now we know. The JK development team gave us a hint about what we can expect at the end of their YouTube festival yesterday. Looks like they'll be using some type of audio relay infrastructure. Proprietary service that JK clients will all connect to that will relay their packets to all other clients in the same session. Maybe using MPLS protocol to provide QOS transmission priority for these packets. Dunno. I'm guessing there is an existing service out there using globally placed hardware that can perform this service. What it will cost hasn't been revealed yet. They claim that it will not only significantly reduce latency but also lower the bandwidth requirement for each user. Sounds like a move from p2p to client-server. This will of course require globally placed hardware with high reliability. RE: Reduction of Internet latency in sessions... - kempjef - 05-31-2020 I'd be glad to pay for a service that solved the latency issues (if latency could be reliably brought down to 10-20 ms for a 300-miles-apart session, for example). The business model for the JamKazam guys is a bit scary, though--there was apparently not a lot of interest in online jamming before Covid-19, and I would guess that most of the current interest will disappear when a vaccine is found, especially for a pay-to-play service. A bit of dilemma there. Musicians don't typically have a lot of money to throw around, virus or no virus. RE: Reduction of Internet latency in sessions... - StuartR - 05-31-2020 (05-31-2020, 10:29 PM)kempjef Wrote: I'd be glad to pay for a service that solved the latency issues (if latency could be reliably brought down to 10-20 ms for a 300-miles-apart session, for example). The business model for the JamKazam guys is a bit scary, though--there was apparently not a lot of interest in online jamming before Covid-19, and I would guess that most of the current interest will disappear when a vaccine is found, especially for a pay-to-play service. A bit of dilemma there. Musicians don't typically have a lot of money to throw around, virus or no virus. RE: Reduction of Internet latency in sessions... - Zlartibartfast - 06-01-2020 Latency along the internet backbone has dropped a lot in the last 5 years, with upgrades to fiber and 10Gig copper all over the country, However, it's the "Last Mile" that's killing it for a lot of people. Regionally distributed servers make sense, but they have to be managed! Adds cost? Improvements in the queing process could be very helpful - 30 - 40 ms latency is not a huge problem if it's the same for everyone! You get much more delay than that playing onstage in a large hall or small arena. Who here has ever played in a marching band? RE: Reduction of Internet latency in sessions... - StuartR - 06-01-2020 (06-01-2020, 04:45 PM)Zlartibartfast Wrote: Latency along the internet backbone has dropped a lot in the last 5 years, with upgrades to fiber and 10Gig copper all over the country, However, it's the "Last Mile" that's killing it for a lot of people. |