Game Discussions Star Citizen Discussion Thread v12

Speaking of coffee cups, Chris stole mine...


Source: https://i.imgur.com/9FX3K3H.jpg
9FX3K3H.jpg
 
I know that i'm wondering what it would cost to build your own backend.

Generally you trade steady ongoing costs for a larger upfront cost but lower monthly costs.

Its all fine and dandy until there are problems, in which case, you need the support staff there to fix it, which is one of the benefits of the cloud. Server breaks, they should spin up new ones for you quickly, and nobody is the wiser (reality might differ from marketing of course).

Now, let me say this straight, i've been anti-cloud since the first marketing demon first said the word, but it does have its advantages.

But if you're going big, then it might be better to roll your own, as economies of scale come into play.
 
Speaking of coffee cups, Chris stole mine...


Source: https://i.imgur.com/9FX3K3H.jpg
9FX3K3H.jpg

You mean under the goblets, or the coffee jug?
Depends which ship, but thought the jug was on the table on the 890; if missing a cup - pretend you're in the Elite Empire, send out a beacon for some players to act as slaves, and sip coffee from a goblet like a King while they fan you :D
To be straight...I haven't had any of the issues plaguing a lot of folks. I've never had an unclaimable ship yet and I haven't had to do a character reset in a desperate attempt to fix it.
Yeah it's obvious some things a character reset wont help with.

My ship bug was just doing a quantum travel, and i "exploded" near the end and straight into a 30K / CTD. Starting up again and ship was unclaimable, maybe because it was destroyed at "Stanton" and not a real place per se?

Apart from that the servers have settled down a lot, but then not played much or how i did in 3.13.
 
You mean under the goblets, or the coffee jug?
Depends which ship, but thought the jug was on the table on the 890; if missing a cup - pretend you're in the Elite Empire, send out a beacon for some players to act as slaves, and sip coffee from a goblet like a King while they fan you :D

Yeah it's obvious some things a character reset wont help with.

My ship bug was just doing a quantum travel, and i "exploded" near the end and straight into a 30K / CTD. Starting up again and ship was unclaimable, maybe because it was destroyed at "Stanton" and not a real place per se?

Apart from that the servers have settled down a lot, but then not played much or how i did in 3.13.

325A, its defiantly missing.

From 2 years ago...

Source: https://youtu.be/Nje2Ws-4ns8?t=129
 
lets see.....

some random numbers

vCPU = 8
memory = 32GB
system = windows
plan = Standard Reserved Instances


20,000 instances = $4,323,200



hmmm, if CIG is running 20,000 instances, that means there are 1 million concurrent users. If they run one game server per EC2 instance.

Interesting.....
A few caveats:

Server pricing varies from location to location. Multiple locations mean multiple server costs per virtual instance.

Not every server will be full. Each server will count as a separate instance for AWS pricing, even if there's just a single player on the server.

CryEngine is notorious about having almost everything done using server side authentication vs client side authentication (yes, it is dumb). This vastly increases CPU usage even when players are performing non-heavy usage tasks.

Faster player or object movement speeds will increase CPU overhead, especially if the players or objects are interacting with the physics engine (you can thank the OG CryEngine guys for this too).

Star Citizen's tick rate is currently set at 10hz (or lower?). It needs to come up by 3 times to meet the demand of most modern twitch based shooters. This will increase CPU cycle overhead drastically -- costs per CPU cycle over your threshold are notoriously expensive. Increasing the tick rate will begin seeing CPU compute cycle costs rise by each hz -- cost per hz will not scale linearly.

And really, you don't have to believe anything I say (and you never should). But it's obvious that CIG could increase performance by upping the server tick rate. They can do this by spending more money on AWS CPU compute cycles.

edit: What you are quoting there is similar to what I run SSMS queries on (using Azure). It's slow enough that my boss complains constantly when we are at heavy usage (in a three person department). You'd need to start using AWS's heavy hitters in order to maintain a constant 30hz per server.

I know that i'm wondering what it would cost to build your own backend.
I'd say you could probably get away with a decent server farm for SC with its current playerbase for around $4m to $8m or so (depending on how many regions you would want to support). This would include initial operating costs, both for hardware and personnel. Monthly you're probably looking at...I dunno, $40k - $80k/m in electricity + server operating expenses + rental space. Staff would be at least another $25k to $40k a month. Switching to your own server infrastructure does leave you waaaaay more vulnerable to security exploits, especially DDoS attacks.

But, going your own way solves two things: keeping your ongoing overhead costs lower than using AWS and getting a better tick rate, which means better performance for your servers and therefore your players.

I'm fairly certain that Turbulent is in the process of dumping AWS for their own internal server hardware.
 
Last edited:
What Ci¬G need to do in the meantime is wipe the persistence DB. They didn't do it prior to implementing baby persistence with 3.8, which they should have done as a matter of course. We haven't had a persistence DB wipe since then...almost 2 years ago.

Due to currently unaddressed exploits, thousands of players are now buying entire fleets of ships in game and currently have aUEC balances in the billions...which is all being handled by the authentication servers and the backend DB services. Ci¬G have made no attempt to curb the exploits at root or wipe the persistence DB to address those issues alone, never mind the other backend DB problems we're all now suffering. It's time to bite the bullet, upset all the trolls using exploits and those buying billions of aUEC on Ebay from those exploiters and just get it done.
I think they are waiting to switch from AWS to their own internal servers for this to happen. Get a new currency and call it bUEC (buck! get it?), fix the exploits, wipe everything out.
 
Last edited:
Generally you trade steady ongoing costs for a larger upfront cost but lower monthly costs.

Its all fine and dandy until there are problems, in which case, you need the support staff there to fix it, which is one of the benefits of the cloud. Server breaks, they should spin up new ones for you quickly, and nobody is the wiser (reality might differ from marketing of course).

Now, let me say this straight, i've been anti-cloud since the first marketing demon first said the word, but it does have its advantages.

But if you're going big, then it might be better to roll your own, as economies of scale come into play.
I've been working in healthcare for almost 15 years now -- AWS and Azure are absolute lifesavers. Mostly because of security reasons -- my old office had to have either a $375,000 security retrofit to continue to host our own internal server...or we could just switch to AWS. Like they were going to have to install security doors, new cameras, a host of crap that cost a ton of money. Meanwhile AWS...has all that security built right in to their farms. Huge benefit, especially if you are working with confidential patient information -- you need to not only have great cybersecurity but also meat-space security..

My old boss complained about the AWS server costs and kept trying to find ways around paying...$1,500 a month. I'm no longer at that company lol
 
Back
Top Bottom