As promised, we put your money to work immediately. Not only have we paid for Inkbunny's main server, but we found a way to use the excess funds in line with our goal of obtaining three years' hosting.
From low-powered backup to secondary server
As noted in our previous FAQ, we planned to upgrade our on-site backup server this year. After all, Inkbunny will soon be larger than 1TB - we just passed 600,000 submissions and 260,000 members.
This configuration also comes with a 100TB/month network allowance, at a speed of 1Gbps. To put that in perspective, Inkbunny currently uses around 12TB/month, on a 100Mbps line.
Compared to our current backup server, it has 4x the raw disk space, 8x CPU performance, 16x RAM, 10x bandwidth, and 20x network allowance - all for just €2003.40 (~US$2400) for three years.
What's it for?
Aside from submission and database backup, we intend to use this server in several new ways: * to offload public content serving, decreasing network and disk utilization on Inkbunny's main server * to test new versions of Inkbunny's software, database, and OS in a near-production environment * to run a warm/hot standby database, decreasing the data-loss window in the event of server failure
Of course, its primary backup role remains important. We don't want to be shipping hard drives in a crisis!
Didn't you say you didn't need another server?
For years, we've maintained an on-site backup server. We took this opportunity to expand its capabilities beyond mere file backups, after evaluating the likely cost of main-server upgrades if we didn't do so.
We've seen increasing use of bulk downloaders, with hours-long bursts above 60Mbps. We consider this legitimate use, but it has the potential to cause network congestion and excessive disk usage. At the same time, we want to offer new high-bandwidth features without saturating our network link.
The new server will help to solve these issues. We can serve public files and thumbnails from it, while using the main server for the actual pages and private content. It should also ease future development.
Weren't you considering a U.S.-based cache?
Yes; we still are. This is a good first step. Once we split the load, it'll be easier to distribute further. We considered leasing one in a separate sale, but the deals weren't as good.
Why this server/configuration? Why not just upgrade bandwidth?
It's a cheap way of getting 100TB/month transit and a 1Gbps port. Upgrading our main server would be nearly as expensive, and we'd "lose" its unmetered 100Mbps connection (worth ~30TB by itself). Why broaden a road when you can build a new one and keep the one you have for the same price?
We considered a less-capable system with 10TB/mo. for €1000 - but this way, we get copious bandwidth and RAM for caching, and a quad-core CPU of the same architecture as our main system, as well as disk space for backups. As a bonus, our host ran out of the E3-1230, so we got a free 200Mhz upgrade.
We got 50% off the list price (in addition to a 30% contract-length discount) because it's a three-year-old model and limited to a maximum of 32Gb RAM and a single CPU. That's fine for our purposes.
It wasn't at all clear whether we'd raise what we needed, but you came through, and more. We now have a solid foundation for future growth, thanks to the generosity of our members.
Many, many thanks to all those who contributed. We'll continue to accept donations to fund our other expenses and future payments. However, we encourage you to consider a commission or two as well!