How self-hosting compares to paid hosting — I was surprised
In this week’s OpenSim Hosting Providers survey, in addition to a list of commercial hosting providers, there was also an option to say that you hosted your grid yourself.
I expected the commercial providers to do better, since they were, after all, the professionals. But the results surprised me.
When it came to stability — whether the OpenSim install crashed a lot or not — the self-hosted respondents scored their grids better than both the Zetamex and the CloudServe customers. Only Dreamland Metaverse did better in this category.
In performance — how fast or laggy the regions were — both Dreamland and Zetamex scored better.
The one result that did not surprise me was that on the interface, since all three commercial hosting vendors have put quite a bit of work into their web-based management panels.
So does this mean that everyone should run out and run their own OpenSim?
- Running an OpenSim server is like running a Web server. Times three. Would you run your own Apache server? Can you set up a MySQL database? Can you forward ports on a router? These things are not impossible, you can learn how to do it — I learned how to do it, and I’m an idiot — but you have to really want to learn.
- Home-based connections are slow. Don’t expect more than a couple of people to be able to visit you at once, unless you rent a server in a commercial data center somewhere.
- Setting up a region or a grid is just the start of it. You have to be able to do your own upgrades and backups. In my case, it was the ongoing maintenance chores that got me to switch to commercial hosting.
Anyway, back to the survey.
I also ran a chart of the individual flavors of OpenSim, but there didn’t seem to be any consistency to the results.
Although WhiteCoreSim scored better in stability than standard OpenSim and the two packaged distributions, it was only used by two of the respondents. It’s the most cutting-edge of the various flavors of OpenSim, and, with a smaller community to draw on for support, probably requires more technical skill on the part of its users. Or maybe the small sample size just makes the results meaningless.
What did surprise me very much was that none of the respondents were using the New World Studio version of OpenSim, possibly because of a lack of support, updates and communication from its creator.
I was also surprised that Sim-on-a-Stick didn’t score better on the interface, since it is easier to set up than the other versions. Meanwhile, it should have scored the same as the Diva Distro for stability and performance since Sim-on-a-Stick is actually based on the Diva Distro. The variation is mostly likely due to variation in the technical skills and hardware configurations of the users.