|
Good points, I should be more descriptive. Jellyfin and co are media centers, they provide the ability to stream media from a hard drive with a nice user interface. This is for someone who wants to run their media server w/ a lot of users, or someone who wants to be able to run a single server and not have to deal with transcoding to different formats. As of right now, it's not "better" than these others since the frontend isn't finished, but when it's done it'll be better in two main ways; able to scale the frontend, able to reduce global latency by using s3 bucket(s), able to have a full e2e approach to media importing (everything is the same format, same subtitles, same layout by default) usually that has to be done manually by whomever operates a server. It's using k8s so that individual microservices can be scaled using the autoscaler microservice. This allows users to use premptible/spot machine types in their workloads and only pay when they are using them, which makes converting cheap. High load is users watching a lot of media, to get S3 to work w/ any alternatives you have to run gcsfuse which is expensive to use and a bottleneck when even two people watch different shows. Normly you'd run two or more instances then to load balance, well, with all of these they use sqlite so there is no load balancing capabilities. Runtime costs are important to anyone who has tried to run a single Plex server in the cloud and had to pay a lot of money to be able to have fast, real-time, on demand transcoding so a user could watch a movie/tv show. Ideally you want to pre encoded all of those, that was the original goal of this project, support that. Now it's turned into a full e2e approach to remove dependency on those media centers that aren't cloud first. I hope that explains it better, it's a little niche, but for anyone handling a lot of users I think it's useful, either way for me it's useful. |