EDIT-I feel the need to clarify that I don't really care either way, I have no ambitions to be an instance admin, I just want to do whatever's best/preferredf or the community-

Hello everyone, I'm making this post because I have received a very enticing offer that I think we as a community should consider.
Ruud from lemmy.world has reached out and offered to set up our own instance under the domain of 196.social, hosted on lemmy.world's hardware. This would allow the community to exist as a separate entity from any one instance, while also improving stability, as it would not be impacted by The same volume of hacks or attacks. It would also not be affected by Lemmy.worlds downtime as it would be fully separate.
The one trade-off is that we would have to fund this ourselves. Ruud says he is more than willing to cover the costs initially until we get some sort of funding set up, but after that grace period we would have to crowdsource the money. He says that the expenses of running a one community instance are very low, however. especially if we don't allow account creation, so it would not take many people to fund it per month.

Please vote in the comments whether we should remain on blahaj.zone, or migrate to our own instance.

you are viewing a single comment's thread
view the rest of the comments

[–] abs_mess@lemmy.blahaj.zone 9 points 1 year ago

I vote stay.

While the Fediverse is built to be decentralized, social networks rely on network effects. People gather where other people are. There will always be 'winners' and 'losers' in building an instance, or even making an alternative 196. c/196 f/196, c/195, fedi196.gay, ect. The original started out with over half a million members, small by Reddit standards. We're the largest alternative around, and we got less than 10k.

Administering an entire instance for one community seems like overkill, especially when there is a path forward to reducing costs significantly.

Also I like the vibes here.

But if you are concerned with server costs...

The existing plan should take care of the significant majority of costs. AWS charges out the ass for data egress, and while I don't know the billing specifics, this instance moves roughly 1TB a day, and the majority of it is static images by virtue of well, images. It was mentioned that this covers half of existing costs upwards of 1-2k. So whats the obvious solution? Don't use AWS for storage or transfer.

Cloudflare and a significant chunk of other providers are part of something called the Bandwidth Alliance (which AWS declined to be a part of) and most importantly, these providers offer S3 compatible object storage without data egress fees. Lemmy uses something called pict-rs; on default settings, uses local storage that has to be part of the VPS that you rent. Enabling object storage allows you to move storage away from the VPS itself and onto a different provider - one that does not charge data egress fees. (Easy money? idk) Ideally, you also get edge caching as a bonus.

So what about the data transfer not associated with images and can't be easily remedied? Different VPS providers have different benefits. Digital Oceans lets you customize your VPS to the Nth degree, Hertzner offers up to 20TB of transfer with it's VPS offerings. While I can't vouch for them myself, that should be enough to cover a whatever other thumbnails and other data that can't be shoved into object storage. Either way, still cheaper than AWS.

As for AWS load balancing, uhhhhh standard nginx/apache config? I've only used nginx as a reverse proxy but I'm not paying for it.

Not touching DB management with a 10 foot pole

TLDR: AWS is great for small projects and my relatives drooling over A40's but they charge for egress, and other QoL services that typically get used by companies with deep pockets.