Videoconference performance

Ok. You try to frame things as if I need to offer an alternative to your current approach, but all I’m saying is that you should stick to what Nextcloud told us was the business model – working with the community to create fully open software without restrictions, and selling professional support, SLAs, and consultancy to large enterprises. I don’t need to propose a new idea – I was quite happy with that one! Originally, it was specifically stated that the target customers were very large enterprises into the thousands of users, not small organisations. This idea that any installation with needs beyond the size of a family unit must be a business that should pay up (and the attitude towards them) is something that’s crept in over time, and I think it’s detrimental to the project.

If you were looking for ways to monetize smaller organisations, I think you made a mistake by not offering your own hosted service. I realise that this might seem counter to the idea of a private cloud, but I don’t really think that makes much sense. In the end, if people trust you to create the software that protects their privacy, they might as well trust you to provide a private hosting service, too. By using third-party hosts we have to trust two parties instead of just one. And, crucially, anyone that chooses not to exercise that trust still has the option of running their own server. To my mind, that is a very acceptable offer from the company to the community – you are free to expend your own resources on supporting your own installation, but we can offer good value on running it for you. I think that really is a win-win, with no need for shenanigans.

In terms of you trying to grow the company to the point where it can single-handedly take on Microsoft, I’m certainly not harbouring that delusion, and I hope you’re not either. But the logic that whatever action is good for the company’s profits is therefore good for the open source project is clearly flawed. One doesn’t have to look very far around the open source landscape to see that. That’s what Owncloud was (allegedly) doing, and then it got forked…

The strategy that might actually make a difference (and the one many of us bought into), is to build up a healthy open source community that works together towards the stated goals, by keeping the needs of the community and the company in balance. I think very few people are going to be interested in contributing to Talk, when clearly the real action is in the HPB. Going down this road of embracing open core functionality, and trying to extract from smaller organisations, is very short-sighted - what you are actually doing is limiting yourself to being a relatively small player. IMO, this is not the way to build the sort of community project that attracts such wide buy-in that it disrupts the industry (and they do exist). I also think that the attitude you’ve taken recently will actually work to limit the potential of Nextcloud in that larger sense, even if your balance sheet has seen a boost.

I’m not suggesting that you rewrite the HPB and OnlyOffice, which is clearly infeasible. The limitations of both Talk and OO for community users have recently both been sharply highlighted. Resources spent on Talk and OnlyOffice could have been spent on integrating Jitsi and Libreoffice Online (which impose no proprietary restrictions) and helping to improve those projects. In terms of suggesting positive ideas, that would be one of them.

In the case of Talk, we must bear in mind that Nextcloud, Spreed, and Struktur AG cannot really be thought of as separate entities. If Struktur/Nextcloud really wanted to stay fully open, they have the option of open-sourcing the HPB tomorrow – no rewriting required. I doubt that’s going to happen though, because it’s all about selling the proprietary product.

In terms of what I’m saying not having any positive purpose or proposed solutions, that’s entirely wrong. I’m saying that there has been a clear change of direction and attitude in recent times, which I think are mistakes for the project, and I would like to see the course corrected to bring the needs of the company and community back into better balance. In particular, this attitude of suspicion and derision towards anyone with needs beyond those of a small household needs to stop.

6 Likes

Well, this user understands that perfectly well. The question at hand is about the strategy of the company in the context of a community/company open source partnership project, and how it affects that relationship and its future. And, also the integration of proprietary software into the mix. Please don’t be simplistic.

3 Likes

Affordable doesn’t mean the same as “free beer”.

Okay, then how about this (call it “Nextcloud Hub for Professionals” if you like):

  • Strictly self support (or maybe one ticket per year for 30 minutes, everything above costs extra)
  • Access to the Nextcloud portal for advanced configuration documentation
  • Onlyoffice with mobile editing and other “advanced” features for 5-10 users
  • Nextcloud Talk with HPB for 5-15 users (maybe 10 users and 5 guests)
  • Outlook Plugin for 5-10 users

Putting this on UCS for example would make it a complete solution from the ground up if it is affordable priced for SMB companies.

1 Like

You can improve your know how. Also you can improve the source code.

1 Like

Let open a real community driven experience and a call for donations and a bounty for features.
Dev will get paid, if not enough money comes in, it will be on the community dev to work on it because Nextcloud GmbH can’t pay dev for that.

This would be similar to OPSI co-funding of features:

As @Semjel said, don’t be simplistic and think people here don’t know what they are talking about.

Yes. But you can also use a microsoft cloud, pay for it and downgrade your know how. And yes you must not read source code and you can never modify software or software packages.

1 Like

Nextcloud Talk Backend As posted here is being coded in linked Gitlab.

Since the default internal signaling was not usable on my system and Nextcloud’s own High-Performance-Backend is not affordable by private users, I decided to take a shot and implement a signaling server based on what can be found in the source of Nextcloud Talk and the API documentation.

Signaling Server

The signaling server itself communicates over a WebSocket connection. For better performance it is recommended to use a separate MCU/SFU. I worked with Janus but I try to keep the interface abstract so it should be possible to expand the connectors for using other servers (Jitsi, etc.)

8 Likes

Hi thanks a lot for this.
I want to test IT.
Is it possible to share a little howto to get this running on Ubuntu 18.04 lts Server?
Best regards Patrick

Two interesting projects that allow to use nextcloud with BBB and jitsi:

2 posts were split to a new topic: Private use of talk during Coronavirus disease 2019

Hello,

In my school, we have big blue button and it works for more than 100 people !
And zoom is not a good solution : problem of security, they sold your data to FB, it’s an proprietary American application (patriot act)…

3 Likes

Thank you so much for this effort! As Nextcloud is not keeping up with their promise to deliver 100% Open Source it is great to see that the community is willing to make up for that.

At the same time, it seems that Jitsi is where the larger communities rally around.

2 Likes

There is now a first release of an app to integrate big blue button:

3 Likes

Mmh, great :heart_eyes: Unfortunately, I dont have admin access to the server BBB. So I cant connect Nextcloud of my school association on the BBB server.

You could see if the school might add a chat bot to call BBB from Talk using this framework.

Is it possible to use this solutions for integrate NC Talk with Jitsi Videobridge as an SFU server?

BTW, thank you

Right now that is only possible through BigBlueButton. See the app here or download in the appstore.

Sorry to re-animate an old discussion, but I have some news on both fronts.

First, the latest versions of Talk include a LOT of performance improvements. Without a back-end, conversations with 5-10 participants should generally be possible (provided enough bandwidth and compute on the clients) . Talk 9 will include more improvements as well as a grid view to make calls with more than 5-6 people also nicer in the user interface.

Second, we continued the conversations with our partner, Struktur, and we are today happy to announce that they have agreed to open source the back-end.

9 Likes

Hi,
do you have any update about talk performance and features? Jitsi now supports both e2ee and VP9 video codec.