🇳🇱
So I've been reviewing our server plan here at QOTO to figure ut an upgrade plan...
The mastodon instance itself runs on a 32 gig memory system that is maxed out and thats costing me about 120/month. The bulk of that cost is the result of our very high federation count (we are the top 20 most federated english servers). We have to process so many external statuses it costs a lot of resources. the advantage means when someone posts your posts federates to the maximum number of public feeds possible, so our users have one of the largest voices in the fediverse. It also means we have a huge database of text searchable toots going back years.
If we want to maintain that level of federation the next step is to double our server costs up to 240$ /month. Thats just for the main mastodon server. Our NextCloud, PeerTube, Gitlab, Funkwhale, and other auxiliary services add somewhere to the tune of 500$ / month on top of that.
I guess this is the price of fame :)
@freemo that is an astronomical cost. and I wouldn't expect you to cover out of pocket.
Reduce something, connections, keep performance better at same or lower costs.
Heck, test drive Pleroma and if we can convert the database it would be a great option.
@design_RG Its worth a try, but very unlikely to make a huge difference (perhaps enough to cut the memory in half and keep the cost at 120, but thats if were lucky).
Almost all the load comes from the fact taht we have such a large federation footprint. Not just in terms of what servers federate with us, but int erms of the individual posts that get sent to us from throughout the fediverse (which allows us to have features like remote timelines). We process several hundred remote interactions per minute, its huge.
Does pleorma even allow for remote timelines? Does it have subscribe vs follow, does it even have the features we are investing in having I wonder?
