Upcoming noteworthy MistServer changes

This gives the ability to track the effect of changes to the setup in the viewership’s ability to watch streams with a good experience. Perhaps the single most important metric of all.

This project started as an internal tool to test our own performance, but we found that many users of MistServer are also interested in getting these kind of metrics for their installs to assist with scaling better.

While this feature might not sound very exciting - and indeed, it does not bring any new capabilities for the average user - it does unlock a few key things for improving MistServer in general. Specifically, this allows us to write unit tests for Mist's various inputs and outputs that can be ran under controlled conditions without having external state that may interfere with them. It also opens the door to processing pipelines, where MistController can convert files or streams between various formats automatically and/or perform operations on them that then have a result that can be stored back to disk. Livestreams already had the ability to run processes like this, but VoD assets do not have this option currently. This change is a big step towards enabling these workflows for VoD as well.

This is another one of those projects that keeps being pushed back, despite being long overdue. While the current web interface is servicable, it’s nothing more than that. Only parts of the current API are used, let alone the new APIv3.

While it is already possible to do this by polling each server and adding the counts together and then inserting them into the player manually, this is a fairly common setup and it makes sense to support it officially. Because the media server itself keeps the count, it also updates instantly and can piggyback on existing connections that are needed to play the stream anyway. This makes a native setup significantly more efficient than it would have been otherwise.

Currently, MistServer has pretty good support for VoD subtitles in various formats, including even some conversions between formats. For livestreams, however, there are quite a few unsuspected caveats and difficulties. For example, there is no clear ingest method to insert them, and even when inserted through some roundabout way, most outputs don’t time/sync (or support at all) live subtitles as well as you’d want/expect them to be.

An upgrade to the API has been long overdue/planned. This item should really not need further explanation; anyone that has worked with the old API in the past will know why this is important.

SNMP is a common method to monitor professional equipment in a network. It doesn’t add anything else beyond the ability to use this standard to access already existing metrics.

MistServer already has (or is about to gain) most of the functionality needed to do this effectively, so it just makes sense to support actually using it like this.

This will allow using MistServer for use cases such as dynamic (and personalized) ad insertion while making it impossible for viewers to block them, running geo-aware (or even user-aware) content, serving high quality for paying customers but low quality after the first few minutes for non-paying customers, switching what a user is watching from the server side without letting the client side change it, etc.

MistServer has had an experimental load balancer since 2016, but it was never developed beyond a proof of concept that “just works”. Not only do we want to make this a “properly” supported feature (as opposed to a standalone tool), but since then we’ve also learned a lot and want to integrate our new learnings as well.

Currently transcoding on the local machine requires using ffmpeg or writing your own script that integrates any other transcoder. This native support for some of the most commonly used hardware accelerated transcoding methods improves ease of use and ease of deployment.

The lack of Windows support in 3.0 was a regression that we accepted due to time constraints, not an intentional change in supported operating systems.

This enables workflows where VoD assets are stored across potentially infinite servers, rather than needing a central storage location. It also allows workflows where directories containing stream data may not be writable for MistServer for any reason (these workflows are currently needlessly slow, since the header file must be re-created every time file access happens).