Developer - The Undermine Journal
Hey folks, I'll try to sum up instead of replying point-by-point.
The issue at Linode is indeed the high number of disk operations required on a speedy SSD. Let me repeat that: it's not disk *space*, anyone can get disk space for cheap, it's high-speed constant disk *activity*. A traditional spinning HDD is not feasible, hasn't been since we started. The majority of our disk activity is maintaining and updating the database, reads are well-cached and aren't as constant or as costly as writes, so it's not "bad traffic" causing problems or anything. TBH our best traffic days were at the beginning of this year; traffic has diminished since then (to be expected at the end of an expansion). The reason it got on Linode's radar now might be because Blizzard is pushing out frequent EU updates (most under 40 minutes apart), tho I can't say for certain since we've been well over their threshold for months.
So requiring SSDs cuts out most VPS providers right there. And any remaining VPS servers won't want to share their SSD between myself and other customers, because while Linode was kind-of a dick with the short notice on this, their infrastructure has always been pretty well set up and if they can't handle it (at a reasonable price), nobody can. Note I said "reasonable price": Amazon could probably do it, but their pricing model scares the crap out of me. So that takes us to dedicated hosting.
I briefly looked at some dedicated host pricing with SSDs, and it's around $200/mo (almost double what I'm paying Linode now). Plus I have no guarantee their hardware would be sufficient for what I need to do, or that the host won't jack up prices or go out of business or whatever. Between subscriptions and ad revenue, I could cover a $200/mo cost usually without a problem, but I'm just not comfortable with that cost for what you get. I don't think it's worth $2400/yr, you know?
If TUJ comes back, it'll be on dedicated server(s) I host at my house. Yeah, it sounds a bit sketchy but it can be reliable and cheap. TUJ's database was hosted at home until 3 months ago anyway. I sold the parts on eBay once Linode was stable for 6 weeks or so.. wish they told me it'd cause a problem sooner otherwise I could've just moved things back here. Anyway, I can build another heavy-duty database for about $1000, and with some parts I already have around I can have a dedicated web server for probably under $200. I'm not concerned about that one-time cost, I'd make it back in subs and ads after a short while. I'd upgrade my home cable service to business-class so it gets a static IP, unblocked ports, and quicker response to any problems that come up. That'd cost me about $60-$80/mo more than my cable bill now, but it'd end up about the same or less than what I'm already paying Linode, so it's basically the same recurring cost as before.
So, problem 1, which is moving off of Linode, isn't really a huge deal. Yes it'd take me time and money to build the system, and yes it's a pain to have to deal with Comcast to upgrade my Internet service, but it's doable. I don't need anyone's money for this, I don't need donations or a kickstarter or a friendly new hosting service, I'd just need to buy the parts, set it up, and make the money back over time. Money is not the problem here.
But there's problem 2: Undermine's code is creaky. It's been on Linode since 2010, and been modified a dozen times to work around limitations in where servers were and what operations were speedy and connected realms and.. it's a mess. The site had to be live all the time so the fixes weren't ever rebuilds, they were just patches. I don't want to try to get that entire tangle out of Linode and onto another box because it's already a mess, and hooking all that up again would be a nightmare.
So, I'd pretty much rebuild Undermine. Yes, I'd probably reuse a lot of functions and routines I have now, but really it'd be a whole new database schema and mostly new code. New site layout, too. I've learned a lot in the past 4 years and the codebase could use a cleanup. But this is the real problem. It'd take lots of time to write all that, and I'm not sure I have the energy to do it. And no, I don't trust others to help out nor do I want to open-source it or any of that. I want lots of control over stuff like this, so people don't get upset looking for a cut of subscription money or arguing about details. I rule TUJ with an iron fist.
So what I need to figure out is, do I have the time and energy to do this? Can it be done in a reasonable amount of time? I need to think it over. Then, if I do it, that'll take a while.
And I won't sell the site to Curse. Or anyone, but especially not Curse. I work for ZAM, why would I want anything to do with Curse? It won't end up on ZAM, either. Like I said, I want to rule TUJ, and I don't want to answer to anyone with my pet project.
edit Damn, that got long. Hope that answers your questions.