when I start writing this comment, the post is 47 minutes old. if I understand the linked page properly, lemmy.world has been functional (all green checkmarks) for the past 10 minutes which is the furthest back the data goes. All the other instances are all green except for lemmy.one which is all red. I am assuming that 47 minutes ago, lemmy.world had red boxes?
Maybe a different link would have explained the point better but I don't really see how a 30 minute (??) server outage during an upgrade is compelling to avoid a large instance. Are you suggesting it's better to use a server whos admins don't upgrade? If not, is there really any size of server that would meaningfully avoid this kind of occasional disruption? Seems to me that the dynamism of the environment will inevitably lead to various problems. That's part of the experience. TBH threadiverse uptime on the whole is pretty impressive for such a ragtag groups of admins and devs.
I have accounts on some smaller servers but they have their drawbacks too. Using a bigger server is more convenient because the people and content is already there. It's easier. I didn't plan to use lemmy.world but I ended up making account there to use sometimes.
I think in a year or so the situation might be different. I see the ideological point and I would like it to be true. Maybe the technology will catch up. I think it would be nice to be able to programmatically seed content, but maybe that would be obnoxious to admins.