LugHost - Keeping things moving like a lugnut, geddit?(name is a work in progress)LugHost is a quick and dirty, limited frills dedicated server hosting solution that intends to just get the job done.
The idea here is to launch quickly with minimal features and then build up over time to eventually become the finest dedicated hosting service Blockland has ever seen. Well, the first and second parts were definitely true. Only time will tell if I can actually create "the finest dedicated hosting service Blockland has ever seen."
One improvement that will be implemented right from the start is hosting on a Linux based server. Hosting solutions in the past used Windows servers, which are statistically more expensive and more prone to crashes and bugs, as well as being slightly harder to maintain. A Linux server allows for stability, easier management, is cheaper, and supports the open source movement.
However, there are three unfortunate catches with LugHost that I have to be honest and list up front:
- I cannot run this service at a loss. I have limited savings, and I cannot justify spending time maintaining a service if I am not turning even the slightest profit. (Sidenote - estimations suggest I won't even be making minimum wage (which is about $8/hr), profit-wise.)
- I have to raise the prices. Services in the past have run $6/month, I must unfortunately raise the prices to $8/month. This is counting the cheaper Linux server.
[0]I will not be using a dedicated server - I'll be using a virtual private server. However, the fact that I can choose precisely how many resources I need on a node (at any given time, growing or shrinking to meet demand) to support the players on the node allows me to slash expenditures dramatically, as opposed to a dedicated server which only allows for a specific set of resources at nearly twice the price.
Fortunately, there is one counter statement which helps handle those two problems: By combining the cheaper server, the dynamics of a VPS, and the ever-so-slightly raised prices, the service will be able to monetarily sustain itself with just
3 clients. This means that two of those dealbreaking catches will not come up for quite some time, judging by community interest.
Enough things - lets talk specs for a second.
The VPS will be purchased as a custom plan from
http://myhosting.com . This company has numerous positive reviews both with regards to reliability and customer service. The most recent negative reviews are from two years ago when they did major upgrades to their infrastructure.
As Blockland servers primarily eat up RAM, I can have a server with the following specs that grows as necessary:
- Disk Space - 40GB (Yes, this is sufficient for 10 average Blockland server installations. The majority of space is taken up by cache.db, which does not take up as much space on servers.
[0]RAM - Variable. 512MB added in for each client, up to a max of 8GB per node. Each server uses roughly 400-600 on average, but can grow to 1.67GB before crashing. I will have 12 clients per server. This is perfectly safe - in order for a server to actually utilize 1.67GB, it has to have at least 50 players and hundreds of thousands of bricks. The odds of a large number of servers on a node
simultaneously reaching 1.67GB are functionally impossible. I have run the numbers on this, and I am absolutely confident they are accurate.
- 8 vCPUs. MyHosting does not specify what kind of processor is used. This should not make any tangible difference - odds are the processor is better than the one in your computer.
[0]OS: Ubuntu Server
The server management system at launch will consist of a very simple setup - it will utilize an extremely straightforward text-based interface over SSH (Secure Shell).
If you haven't heard of SSH before, that's fine - I'll explain in a second, but before you turn away take a look at how simple this is:
Look familiar? That's because it's just a simple console - it's integrated, almost exactly like running the server on your own computer. No nasty java interfaces that look like they were made in the 1900s, no web interfaces that
must be ran in a resource-intensive browser - just you and your server.
If your server isn't running, you'll see this screen:
If it is running, you simply see your already-running server (as pictured above).
So why use SSH? There are a few primary reasons:
Firstly, it's a lot easier on me. Managing simply a Linux server doing Linux-y things is very straightforward if you have the right knowledge and experience. I have two years of experience doing just that.
Secondly,
it's secure. As the connection is encrypted end to end, no one in the middle will be able to read or take over your server. Admittedly, this wasn't a very big problem before, but it is
not a problem any longer.
Thirdly,
SSH is completely cross-platform. There are SSH clients for Windows, Mac, Linux, Web browsers, Android, iOS, and probably even blackberries.
Finally,
convenience. Yes, convenience. Option 4 on the menu screen is a fancy way of saying "speed up logins." I'll have a more detailed guide later, but imagine just double clicking an icon on your desktop and your server console just right there - no passwords, no logins, no keys on the keyboard, it's just
there.
In an effort to get this service launched as soon as possible (to fill the gap in hosting solutions), I'm launching it with add-on management via FTP. However, as I understand this is a deal-breaker for many of you after RTB's web-based system, this is priority #1 for new features.
I also intend to have a direct-download-and-install system that ties into BAM's add-on repositories. Until BAM is released, however, I can use a local cache of the RTB add-ons for hyperfast and easy installation.
Also planned, but low priority, is a method for "offsite" logs. Really just outside of the Blockland folder. This means you can give eval to your admins, log it "offsite," and they can't touch the logs to clean up after their vile handiwork.
A lot of you will recall
my previous post on this matter. As you can see, 27 people were interested in a simple, quick solution. I ask you now - now that you know more specific information, how many are still interested?
Here's the tentative schedule (in terms of order of events, not dates) (filled in dots are completed)
- Poll community for interest - Interested.
- Research feasibility of Linux-based hosting solution: Success on two fronts.
- Put together specific information
- Post this topic you are now reading.
If the community is still interested at this point, at least 4 people are ready to do this thing...
- Finalize construction of Blockland server management system
- Test everything - I can reliably test everything without actually buying the VPS due to owning numerous personal Linux-based systems.
- Collect first months payments from first ~4 people
- Purchase VPS
- Upload/configure everything, and test for a few days with first ~4 clients - live beta if you will.
[0]Add some more people and test dynamic VPS resources (shouldn't take too long)
- Open fully to public.
The bottom line (hurr)
- VPS-hosted Blockland server hosting service for $8 a month
[0]Cannot go
in the red.
- Possibly ready in ~1-2 days (Figure almost entirely pulled out of my ass, really as soon as I can clear a day out of my schedule. That means Tuesday the 25th or Wednesday the 26th this week, as soon as I get at least 4 people lined up.)
[0]Bullet points are awesome.
tl;dr: Yes, it's a lot to read, but I'm laying out the specifics here. At least skim the whole thing over before commenting.
