Dedicated machine price reduction

News for the main page
User avatar
This is my homepage
This is my homepage
Posts: 2389
Joined: Fri May 18, 2012 3:20 pm

Re: Dedicated machine price reduction

Post by soja » Sun Jul 26, 2015 11:03 am

stickz wrote:
soja wrote: So you have a dedi with 4 servers, what about expansion, future servers? If you have a gaming community, people constantly come to you with new server ideas. If you're already maxing out your ram(400MB is a pretty conservative value)
I don't think that's a really conservative value. 4gb leaves plenty of room for expansion in the future with less than half used. Not every server needs tons of ram, some just need CPU to process entities.

Code: Select all

400mb x 4 + os (75mb) = 1675mb.
Like I said, you are basing this in a perfect world. In the real world, you don't purchase a dedi for 4 servers.

EDIT: I would really like to end this argument. If you don't find value in NFO's services, there are other options out there that meet your needs.
Not a NFO employee

User avatar
Posts: 12402
Joined: Thu Apr 18, 2002 11:04 pm
Location: Seattle

Re: Dedicated machine price reduction

Post by Edge100x » Sun Jul 26, 2015 11:14 am

stickz wrote:Well, there's two things you do not offer.

[*] i7-4790k's which is stronger than your best dedicated offering by 8%.
Yes, we already talked about why that is (low demand, the fact that it's a less-reliable consumer part, and the fact that it does not support ECC).

Demand is important because otherwise machines aren't reusable; we lose money when machines sit idle.

Reliability is important to our customers and to us because downtime is a serious issue, facilities are remote, and hardware replacements are expensive. We've seen many problems that ECC has prevented and I've heard quite a bit of feedback from customers about other hosts using commodity parts having frequent hardware breakdowns, so reliability isn't a theoretical concern. (We also used to use consumer-oriented parts, and overclocked parts, here, many years ago, and found them to break down much more often ourselves.)
[*] Streamlined specifications for reduced costs.
As I pointed out before, having more available configurations increases our costs (significantly) because more configurations leads to a higher degree of wasted (idle) machines, and low-end machines are less future-proof. We intentionally and carefully balance what we offer to keep overall costs low while also providing configurations that the majority of customers can use.
The day source engine servers use x8 (400mb) more ram is when the extra ram will do more than eat money.
Spending an extra $50-$100 on RAM up-front (which translates to maybe $10 a month in extra cost for the customer) makes sense in the long-term because we know that many customers already use a significant fraction of this and memory usage for applications is only on the rise. Current games commonly use 500-1000MB per instance and customers who purchase dedicated servers regularly run a dozen or more servers on each machine. Some games that are rising in popularity (such as Rust, 7dtd, and other Unity games) use 1-3 GB per process. Some new games use even more; last year, would you have thought that Ark would show up, with its initial 5-6 GB of RAM usage?

On top of reusability for customers, we need for our dedicated machines to be convertible to regular game server use in our system. There is no way that I'd want a new machine in our system with less than 16 GB right now, at the absolute bare minimum, and even 32 GB is sometimes low. I know that even if that RAM isn't used for game servers directly, it is put to good use by the system as a disk cache.
And is it against terms of service to use 30TB of bandwidth as a migration farm to deflect ddos attacks on your website?
I don't know what you mean by "migration farm", but when we filter a DDoS attack for a customer, the filtered bandwidth doesn't even count towards the customer's allotment.
You're paying for that extra bandwidth and there's no way game servers will ever come close to using it all up.
We use a high number for the included bandwidth because we don't want customers to have to worry about going over. We have raised the number over time specifically in response to customers' bandwidth usage increasing.

stickz, your comments show that you have a relatively narrow view that is defined by your own experience, and not the broader that we have, which is formed by what we and our customers more broadly need and deep, deliberate consideration about how to keep costs down and quality high over the long term (informed by operational experience). As I previously noted, your questions and concerns are not new to us and we have put a significant amount of thought into them over the years; we have also tried things like different configurations and different bandwidth amounts and learned our lessons from those experiences.

I agree that this thread should end. stickz, you've spoken your mind about not liking anything that we do, and I've spoken my mind about our choices, and it's all rehashing at this point. You're not being convinced, and you're not convincing us, and I don't think that will change.

what ever
New to forums
New to forums
Posts: 9
Joined: Wed Jul 22, 2015 6:21 pm

Re: Dedicated machine price reduction

Post by what ever » Thu Sep 10, 2015 11:34 am

I don'y know why you said NFO's price is expensive.
I have been doing reseach all around the internet.


I think you are just some idots that is just trying to make people leave NFO and find someone else. This will never ever happened.

I got two VDS in two location which is at LA and Chicago both are 4 core managed vds. I got a 400+ people community and i never try to show off by it.

Post Reply