• Please make sure you are familiar with the forum rules. You can find them here: https://forums.tripwireinteractive.com/index.php?threads/forum-rules.2334636/

Server Server Provider Performance

Waffenator

Grizzled Veteran
Feb 6, 2011
48
3
Sweden
1. I will not name companies that are performing poorly that the 116th is still trying to work worth. Those that are providing what was promised will be duly named. We cannot speak from everyone's experience but we can speak from our own.

2. London Server

From the start Killer Creation has been an outstanding host. They would not let us rent a 64 slot server to begin with because they made it clear their infrastructure would not support it, but that once upgraded they would offer it to us. That honesty and integrity is critical in a server host.

Killer Creation has does lots of small things that make a big difference. At setup their responses are seemingly automated allowing for very quick setup and basically no wait times. They have been willing to work with us for small issues and gave us the extra 14 slots free for a few days to help them and us decide is their new servers handle a 64 slot server well.

Stress Test. So far the stress test has been going on for about 18 hours with 64 players going for several hours at a time. So far no crashes to report. At a 64 player load ping averages about 100 however, it is stable with no spikes. Having an auto ping kick feature would be of great help in the admin panel as we often have connections from around the world and have to remove the high ping players.

2. Chicago Server

The server provider, shown on tripwire's RO2 server list, has been quite a pain for the 116th. The server, was sold to us as a 64 slot server, we throttled it back to 50 slots, yet still were having as many as 8 crashes a day, and sometimes those crashes would not even enable an auto-restart. They should have not offered 64 slots to begin with.

The server provider had to manually set up the authorization emails to get us started. Their customer service is handled by many agents that do not take the time to read the whole backlog of open tickets or forget what promises they have made earlier and having to be reminded (yes, from the same account agent). They often blame the game or other issues rather than accept responsibility for their own failings. Since we have multiple servers running its pretty easy to catch them in their lies.

Their efforts to satisfy the customer have been lackluster rather than wowing. That is, the combination of not admitting fault and not bending over backwards to say "We want you as a customer" would cause most customers to pack up and leave.

They have promised us that they will deliver a new box "early this week" that will meet our expectations in both horsepower and bandwidth.

The ONLY reason we are sticking with them is to get a clear assessment if they were not simply overwhelmed by the unexpected popularity by RO2 and have a bad egg or two in the bunch....or see if they are truly rotten and the community should be forewarned forevermore not to do business with them.

If the case is the latter, which we do not know yet, then it begs the question...Why did Tripwire choose them as a favored vendor?

On the other hand, it may be that killercreation is simply a gem in the rough an d our US host is the norm?
 
  • Like
Reactions: Norse Soldat
Great post! I did enjoy playing on the London server. I have not tried the Chicago server and doubt the ping times would allow me to play well anyways. I totally agree about the stability of the London server and because in game glitches have experienced zero issues.

Report as of yesterday:

Earlier in the day played a few different times throughout the day I maintained a 80ms-90ms Ping time coming from Iraq. This was when it was at 50 slots and It was close to full or full the whole time.

When it was bumped to 64 slots the ping elevated for me to 100ms-120ms, very playable still (especially being where I am). I am definitely not afforded any advantages with this ping but it is constant and my play style will adjust. Right now I suck : )

Look forward to slaughtering some more allies soon!

P.S. As far as the other host I believe in voting with your dollars, be very clear that this is their last chance and then you are out!
 
Upvote 0
Average 100 ping is too much, the server is overstressed.

You might want to do a bit of research or at least come up with an reasonable explanation before you start to spout pish like that.

I have been monitoring our server constantly while playing and looking at pings - Which has quite often hit 100 for quite a number of players, but the server software has only been using 25% CPU usage on one core and 400MB of ram out of an available 2gb to the game.

The Network utilization has been less than 1% over the period of having a full server 38/38 (We limited it to 38 for initial performance concerns however this would appear to be too limiting at the moment.)

For Some reason quite a few servers are reporting that the pings claim for no apparent reason - It certainly not down to the server being "overstressed".

It is also worth noting that if the server was overstressed other applications within it - or at least the same core - would be affected. However none of them are - Teamspeak has no packet loss due to processing difficulties and the web server - and other services - are still performing well. Meaning it is something with the game server which is not 100% optimized causing pings to rise to around 100 as the server gets to its peek usage level.
 
Upvote 0
You might want to do a bit of research or at least come up with an reasonable explanation before you start to spout pish like that.

I have been monitoring our server constantly while playing and looking at pings - Which has quite often hit 100 for quite a number of players, but the server software has only been using 25% CPU usage on one core and 400MB of ram out of an available 2gb to the game.

The Network utilization has been less than 1% over the period of having a full server 38/38 (We limited it to 38 for initial performance concerns however this would appear to be too limiting at the moment.)

For Some reason quite a few servers are reporting that the pings claim for no apparent reason - It certainly not down to the server being "overstressed".

It is also worth noting that if the server was overstressed other applications within it - or at least the same core - would be affected. However none of them are - Teamspeak has no packet loss due to processing difficulties and the web server - and other services - are still performing well. Meaning it is something with the game server which is not 100% optimized causing pings to rise to around 100 as the server gets to its peek usage level.

Thanks for the information it is very valuable to the 116th and surely the greater RO2 community. Salute!
 
Upvote 0
You might want to do a bit of research or at least come up with an reasonable explanation before you start to spout pish like that.

I have been monitoring our server constantly while playing and looking at pings - Which has quite often hit 100 for quite a number of players, but the server software has only been using 25% CPU usage on one core and 400MB of ram out of an available 2gb to the game.

The Network utilization has been less than 1% over the period of having a full server 38/38 (We limited it to 38 for initial performance concerns however this would appear to be too limiting at the moment.)

For Some reason quite a few servers are reporting that the pings claim for no apparent reason - It certainly not down to the server being "overstressed".

It is also worth noting that if the server was overstressed other applications within it - or at least the same core - would be affected. However none of them are - Teamspeak has no packet loss due to processing difficulties and the web server - and other services - are still performing well. Meaning it is something with the game server which is not 100% optimized causing pings to rise to around 100 as the server gets to its peek usage level.
Well I couldnt know that since I havent got access to the server:p And you cant either.

But you know what I mean, if the ping in-game is considerably higher than it should, somethings not right, e.g. the server is overstressed.

EDIT: Also it seems like the same problems on the server side have continued from ROOST to HoS, too many servers which are overstressed and have either huge pings, packet loss or at worst, both.
 
Last edited:
Upvote 0
Not really. The pings in RO2 are vastly different from most other games. I suspect it is reporting round-robin values.

As an example, when I run Arma 2 on our box (on an i5-2400), my ping to it is 17ms. Doing a ping from cmd.exe gives the same. Average player ping is about 50ms.

When the Arma 2 server is using all the available resources it can (almost two cores of our four, with one completely maxed out) my ping is still never above 50 and rarely above 30. Average player pings do not go above 70. Actually, I'm not entirely convinced the pings change even this much.

When I'm the ONLY person on the RO2 server and nothing else on the box is in use (or even running), my ping is 60. CPU usage is 0%. When we fill out the server (50 slots), cpu usage peaks at about 20% (80% of a single core) and my ping doesn't really increase. Average player ping tends to be in the 60-80 range with a couple of outliers around 150 which are more to do with where people are connecting from, as they are consistant.

When we had 64 players on gumrak (tanks require more cpu) as a test, pings shot up to the 250+ range and the lag was obvious.

There are also regular spikes where the game has to calculate building damage and artillery strikes. These are, IME, more of a server performance issue than the size of the maps themselves, and can vary the cpu usage by 20%+, which is why we don't increase our slot count to 60 or something.
 
Upvote 0
Well I couldnt know that since I havent got access to the server:p And you cant either.

But you know what I mean, if the ping in-game is considerably higher than it should, somethings not right, e.g. the server is overstressed.

EDIT: Also it seems like the same problems on the server side have continued from ROOST to HoS, too many servers which are overstressed and have either huge pings, packet loss or at worst, both.

1892_mad_smiley_banging_head_against_a_brick_wall.gif
 
Upvote 0
Not really. The pings in RO2 are vastly different from most other games. I suspect it is reporting round-robin values.

As an example, when I run Arma 2 on our box (on an i5-2400), my ping to it is 17ms. Doing a ping from cmd.exe gives the same. Average player ping is about 50ms.

When the Arma 2 server is using all the available resources it can (almost two cores of our four, with one completely maxed out) my ping is still never above 50 and rarely above 30. Average player pings do not go above 70. Actually, I'm not entirely convinced the pings change even this much.

When I'm the ONLY person on the RO2 server and nothing else on the box is in use (or even running), my ping is 60. CPU usage is 0%. When we fill out the server (50 slots), cpu usage peaks at about 20% (80% of a single core) and my ping doesn't really increase. Average player ping tends to be in the 60-80 range with a couple of outliers around 150 which are more to do with where people are connecting from, as they are consistant.

When we had 64 players on gumrak (tanks require more cpu) as a test, pings shot up to the 250+ range and the lag was obvious.

There are also regular spikes where the game has to calculate building damage and artillery strikes. These are, IME, more of a server performance issue than the size of the maps themselves, and can vary the cpu usage by 20%+, which is why we don't increase our slot count to 60 or something.
Yes, I do know that.

I meant too high ping as in, if I play on the server with 0 players my ping is 30 and server full my ping is 65, or 80 or 100 etc etc..

And about packet loss, there were some servers in ROOST that had continuously packet loss and I have found some in RO2 as well, this means everythings definitely not right with the server, there should never be any packet loss.
 
Last edited:
Upvote 0
Yes, I do know that.

I meant too high ping as in, if I play on the server with 0 players my ping is 30 and server full my ping is 65, or 80 or 100 etc etc..

And about packet loss, there were some servers in ROOST that had continuously packet loss and I have found some in RO2 as well, this means everythings definitely not right with the server, there should never be any packet loss.

Your ping will rise as more people connect, that is normal for any game however UE shows pings in a different way.

What you described is 100% normal :)
 
Upvote 0
From Micah at evenbalance regarding pings:

The way ping is calcuated in UnReal is a bit strange. It's not the same thing that you and I think of as standard ICMP. With PB enabled, there is slightly more traffic between the client and server. This may show an increase in the game ping (20 wouldn't be unreasonable), but that doesnt mean everyone will be suddenly lagging or rubber banding.

The guy works closely with the engine and knows it better than we do.
 
Upvote 0
From Micah at evenbalance regarding pings:

The way ping is calcuated in UnReal is a bit strange. It's not the same thing that you and I think of as standard ICMP. With PB enabled, there is slightly more traffic between the client and server. This may show an increase in the game ping (20 wouldn't be unreasonable), but that doesnt mean everyone will be suddenly lagging or rubber banding.

The guy works closely with the engine and knows it better than we do.
Thats good info.

But still I run into over-stressed servers too often :(
 
Upvote 0
Actually Punkbuster should NOT make your pings go up 20ms, and should have a negligible affect on ping. If you are experiencing a consistent ping increase of 20ms when PB is enabled vs disabled then I would like to know and will bring it up with the PB guys.

Regarding ping in Unreal - the way that Unreal calculates the in-game ping is different than some other game engines. Some other game engines split out their network updating from their gameplay code and even if their gameplay code is running slowly due to CPU load, the game will tell you your ping is the same. Unreal actually gives you a ping that represents how long it is taking the server to process the gameplay code. I think this is really a truer representation of what is happening on a server. But you will see your apparent ping go up as a servers CPU gets under load. This is usually fine as long as the servers CPU isn't overloaded. On my test servers with 64 players my ping will go up about 40ms and will be in the 75-85ms range (the server is located within 50 miles of our office).
 
Upvote 0
Actually Punkbuster should NOT make your pings go up 20ms, and should have a negligible affect on ping. If you are experiencing a consistent ping increase of 20ms when PB is enabled vs disabled then I would like to know and will bring it up with the PB guys.

Regarding ping in Unreal - the way that Unreal calculates the in-game ping is different than some other game engines. Some other game engines split out their network updating from their gameplay code and even if their gameplay code is running slowly due to CPU load, the game will tell you your ping is the same. Unreal actually gives you a ping that represents how long it is taking the server to process the gameplay code. I think this is really a truer representation of what is happening on a server. But you will see your apparent ping go up as a servers CPU gets under load. This is usually fine as long as the servers CPU isn't overloaded. On my test servers with 64 players my ping will go up about 40ms and will be in the 75-85ms range (the server is located within 50 miles of our office).

Weird because that came from Micah you put me in contact with regarding pb issues. There most certainly is performance overhead from PB.
 
Upvote 0
Actually Punkbuster should NOT make your pings go up 20ms, and should have a negligible affect on ping. If you are experiencing a consistent ping increase of 20ms when PB is enabled vs disabled then I would like to know and will bring it up with the PB guys.

Regarding ping in Unreal - the way that Unreal calculates the in-game ping is different than some other game engines. Some other game engines split out their network updating from their gameplay code and even if their gameplay code is running slowly due to CPU load, the game will tell you your ping is the same. Unreal actually gives you a ping that represents how long it is taking the server to process the gameplay code. I think this is really a truer representation of what is happening on a server. But you will see your apparent ping go up as a servers CPU gets under load. This is usually fine as long as the servers CPU isn't overloaded. On my test servers with 64 players my ping will go up about 40ms and will be in the 75-85ms range (the server is located within 50 miles of our office).

Thanks for the update. S!
 
Upvote 0
I believe this article is still valid when discussing ping, tickrate, netspeed etc in Unreal based games.

The ping you see in the scoreboard = traveltime from computer to server + 1000ms/tickrate + traveltime from server to computer.

If the serverload is to great for the server to maintain the tickrate, the tickrate will decrease and that in turn increases the ping.
 
Upvote 0