Up- download speeds..

[Kep Kelagin 1]

Hiya all,

I got a new internet connection; ADSL with an up-download speed of 256kilobit-1024kilobit connection. Can anyone tell me how many players i should be able to host? Thx

[Mister Frag 0]

It also depends on how fast your systems is, and how much memory it has, but you should be able to do four or five with VoIP, more if you don't need voice communications.

[Kep Kelagin 1]

tonoight i already had 10 players, without any lag:-)

[RN Malboeuf 12]

Revised for 1.40 net Code

based on Dedicated server @ 1 ghz with 512 megs ram and no AI

if using more then 30 AI deduct 10% as this may lead to CPU slowage

(Note: these figures are NOT exact in any way, but can be used as a guide line)

To check your up/dload band widths go to these sites)

www.bandwidthplace.com/

or

www.pcpitstop.com/internet/Bandwidth.asp

or

www.dslreports.com/

Server Limits (based on Dedicated)

down/up (transfer rates)

A) 56k low end 1.3/1.3 (old phone lines can only handle 2.6k no moatter what way it's going)

B) 56k med 2.6/2.6 (phone line can handle 5.6k but crap modem can't or vesa versa)

C)56k high end 5.6+/5.6+ (can even reach 12 kb locally if you have new phone lines made for data transfer)

down/up (transfer rates)(playerX based on 0-100 ping)

D) Crap DSL 250/50 (4 players max)

E) DSL Low 450/70-100 (10 players max)

F) DLS Mid end 850/150 (12 players +)

G) DSL High end 1500+/100-150 (12 players +)

H)DSL High and FAST 1700+/200-350 (20 players+)

Cable

(note the larger the cites the more node sharing, my node has 3 users on it casue we're a small city, bigger ones can get 30+ users per node and peak usage suffers greatly)

down/up (transfer rates)(playerX based on 0-100 ping)

I) Cable 300-500/10-50 (2-4 players max)

J) Cable 500-750/10-100+ (2-10 players - depends on usage)

K) Cable 750-1000/50-100 (4-10 players)

(note cable below is rare and has less users per node)

L) Cable 1000-1500/100-150 (8-12 Players)

M) Cable 1500-1750/100-150+ (8-12+ Players)

N) Cable 1700+/200-350+ (20+ Players)

(T1/T3 are based on them NOT having any Capped modem restrictions, just cause you have one dooes not mean it runs that fast)

down/up (transfer rates)(playerX based on 0-100 ping)

O) T1 Capped 1500+/100-500+ (10-20+ players)

P) T1 1500+/1500+ (32+ players)

Q) T3 Capped 4500+/1000+ (50+)

R) T3 4500+/4500+ (150+ players) sweeettttt

The number are not exact, but they are close, AND when I say MAX PLAYERS and some one says "nooooo I can host more", I know you can but these numbers are based on having little if any lag

past those numbers and your clients will start to see lag affects in different increments

[MugHug 0]

Just some info for consideration.

I run a 24/7 DS using w2K Server, AMD K7 900Mhz, 500Mb PC133, etc over a 1.5Mb DL/384Kb UP connection.

Using bandwidth testing at the above sites, we seem to have an excellent connection.

Based on the numbers in the post above, it is clear that this cannot answer the original question by its self as evidence indicates other factors involved.

In our experience we have found that a good connection is only the start of working out what a certain DS can handle in the number of players.

One must consider the following factors essential in running a decent DS.

1) The ping of your clients.

a) We find that many people will try to connect and wait in the pool with crazy high pings. Careful monitoring of this and asking the person to leave or kicking them helps a lot.

b) When a client joins your server, it seems to increase everyones ping dramatically untill they have fully connected.

c) One bad ping slows down everyone else. For example, I am 3 feet away from the server on a LAN, but my ping and lag will go up in this situation, so it must be worse for others connecting from outside.

d) Due to experience we limit the max ping to 250ms.

2) How many AIs in a mission?

a) This has a real world effect on use of your DS bandwidth.

b) MP maps with just humans seem to perform better.

3) Are helos, tanks, jeeps and aircraft involved?

a) We find that the use of these can drop the FPS right down on the same exact map.

b) Try to avoid people pairing up in the same helo or tank as this can cause major problems.

c) By default, the Maruaders will normally only have one person in a helo at a time.

d) Multiple helos or aircraft taking off at the same time can cause major problems.

4) The size of the battle area.

a) The further apart players are on the map seems to increase pings and lag.

5) The type of map and number of objects involved.

a) If the map is complex or has too many objects, this can slow things down.

b) Simple maps with players only on foot can keep our FPS above 47 most of the time.

6) The state of the connection between the DS and your clients.

a) A very important point is that anywhere on the connection beyween the DS and your clients can affect the performance.

b) An example is last night. In the early evening the same players used certain missions and our FPS was very good. Late at night for no known reason, the same players on the same missions caused the FPS to drop right down.

7) In-game voice communications.

a) This seems to use a lot of bandwidth.

b) We use a independant alternative on a different IP which dramatically helps.

c) We find Roger Wilco or TeamSound to be the best.

8) The connection bandwidth must be used just for the OFP DS.

a) I run our server on a small LAN at my home. If the wife gets on her PC and uses the Internet, this can cause issues.

9) Designate your DS Server just to run OFP. Run only essential foreground and background tasks.

Our recommedations:

1) The design of the mission plays a real factor in how many players the DS will handle. Avoid eye-candy!!!. So many maps we get are wonderful in design but suck in use under load on a DS.

2) Map designers must test their maps on a live DS under load. To test it just on your workstation will give no indication of how it will perform in MP.

3) Remove all un-essential objects, etc. from your maps.

4) Avoid your wonderful soundtacks as this just adds to bandwidth needs.

5) Use the #monitor command to learn how your server performs under certain situations. Firstly, you can identify a high pinger connecting and kick him before it cuases problems.

6) The #monitor command and the DS info from BIS now gives us some real tools to adjust server config file setting to optimize things.

7) If OFP can use up to 256Mb of memory, then it makes sense that 512Mb would be a good minimum limit in your server to accomodate the O/S, etc.

8) Trying to run a OFP server from the same copy of OFP and on the same workstation where you play is not a good idea because all the graphical work involved will rob your CPU of power and leave little to be used by OFP MP.

9) Avoid clients with pings above 250ms.

10) People in the pool/waiting room during a game can increase lagg and pings!!!

and finally, do not just look at your connection speed. Look at the overall picture and learn how to run your server to the best of its abilities!!!

There are many other factors to consider that are not detailed here, but hopefully this info will help someone out.

What would be nice is a decent FAQ on the day to day running of a OFP DS based on solid experience from existing DS providers.

Regards,

MugHug

(MugHug's Marauders - OFP League CTF Winners)

www.mughug.net

(Edited by MugHug at 10:16 am on Dec. 23, 2001)

(Edited by MugHug at 10:23 am on Dec. 23, 2001)

[RN Malboeuf 12]

>>>>>1) The ping of your clients.

as I stated it was based on a ping of 100

>>>>>a) We find that many people will try to connect and wait in the pool with crazy high pings. Careful monitoring of this and asking the person to leave or kicking them helps a lot.

This does not affect any one playing in the server since the server really does not send any one in the lobby more then a status and text chat

>>>>>>b) When a client joins your server, it seems to increase everyones ping dramatically untill they have fully connected.

Again this is false, the only instant this is true is when the force conect command is given, every one on the server will see a skipp in placement and then some one contects, it happens so rarely that we pin pointed to the force connect comend given with the ASE, this does happens only about 2% of the ppl actually conecting

>>>>>c) One bad ping slows down everyone else. For example, I am 3 feet away from the server on a LAN, but my ping and lag will go up in this situation, so it must be worse for others connecting from outside.

This will have more to do with your net work set up, it sounds like you have a bottle neck some where, countless server never get this problem, we see ppl conect with 400-700 ping on 56 k from the other side of the globe and it does not efftect the server in any way

you see higher pings for everbody only during a map dload from the server

>>>>>>>2) How many AIs in a mission?

a) This has a real world effect on use of your DS bandwidth.

b) MP maps with just humans seem to perform better.

actually there is a formula for this, for every Ai or human  the information increases regardless of what they are, then you take the size of the info packet and multipy that by the number of conected clients

example

0.5kbs info per AI/human (0.5 kbs is an example)

30 AI + 10 players = 40 placement/actions per packet

10 clienets

formula

0.5X40=20kb

20kbs X 10 = 200kbs

as you can see this is a little high, only BIS know amount of info for each action for eact AI/player but as you add more AI and clients the info sent out increases dramitcally

if we knew the precise amount of info we could esimate a better band width usage

<<<<<<3) Are helos, tanks, jeeps and aircraft involved?

a) We find that the use of these can drop the FPS right down on the same exact map.

only if AI are involded, when they are not moving/used they are no different then a placed objected, once used the info is the same amount as it would be for the player

we have played on maps with 30 players and 40-60 vehicles and never seen any more lag then with a 30 player only session

it's more likeley the exploisions they cause can cause frame lag on the clients computer in turn slowing down client updates

<<<<b) Try to avoid people pairing up in the same helo or tank as this can cause major problems.

Totaly False

>>>>>c) By default, the Maruaders will normally only have one person in a helo at a time.

??

>>>>>d) Multiple helos or aircraft taking off at the same time can cause major problems.

in ealer version of the net code this was true, with 1.4 testing we saw 5 Heli take off on east and 6 on  west at the start of the map no such problems

<<<<4) The size of the battle area.

this is untrue sence the data being sent is just location cordinates and size just means a differ number in the info packet

<<<<<<<a) The further apart players are on the map seems to increase pings and lag.

Totaly False

>>>>>>>5) The type of map and number of objects involved.

a) If the map is complex or has too many objects, this can slow things down.

Very true

>>>>>>>>b) Simple maps with players only on foot can keep our FPS above 47 most of the time.

this does not lower ping, it just means the packets sent are not as large and does not increase PING

Ping is the time between two locations, if the server is getting past it's up load limit, and the client is getting past his dload limit will be the only times Pings increase, this is beacuse there are packest being sent before the ping packet and can cause the ping to increase, you shall see this when maps are being uploaded

<<<<<<<6) The state of the connection between the DS and your clients.

a) A very important point is that anywhere on the connection beyween the DS and your clients can affect the performance.

b) An example is last night. In the early evening the same players used certain missions and our FPS was very good. Late at night for no known reason, the same players on the same missions caused the FPS to drop right down.

state? there are many, each client has one between the server, it will not effect the other players

it may be one of the 1st few nodes the server conects to the internet on may have congestion at certin times, in case of T1/T3 this does not appy much

>>>>>>>7) In-game voice communications.

a) This seems to use a lot of bandwidth.

b) We use a independant alternative on a different IP which dramatically helps.

c) We find Roger Wilco or TeamSound to be the best.

100% TRUE DO NOT USE IN GAME VOICE

>>>>>>8) The connection bandwidth must be used just for the OFP DS.

a) I run our server on a small LAN at my home. If the wife gets on her PC and uses the Internet, this can cause issues.

Again based on your network and bottle necks and does not appy to many servers

>>>>>>>9) Designate your DS Server just to run OFP. Run only essential foreground and background tasks.

100% true, dont us your Server to browse the net or play MP3 lol

>>>>>>3) Remove all un-essential objects, etc. from your maps.

100% true, i cant stand seeing 15 vehicles locked

>>>>>4) Avoid your wonderful soundtacks as this just adds to bandwidth needs.

Not true, once the map is dloaded the music is played from the clients copy and does not effect game play what so ever

>>>>>>>5) Use the #monitor command to learn how your server performs under certain situations. Firstly, you can identify a high pinger connecting and kick him before it causes problems.

only if you want to be rude, even if they are playing, they will be the only ppl you see lag during OFp with earlier net code (1.30 and down)

we had 2 700ms ping players these last two days and game play was awesome with 1.4

>>>>>>>>8) Trying to run a OFP server from the same copy of OFP and on the same workstation where you play is not a good idea because all the graphical work involved will rob your CPU of power and leave little to be used by OFP MP.

100% true

>>>>>9) Avoid clients with pings above 250ms.

False now that we have 1.4

>>>>>>10) People in the pool/waiting room during a game can increase lagg and pings!!!

Totaly False

>>>>>>>>and finally, do not just look at your connection speed. Look at the overall picture and learn how to run your server to the best of its abilities!!!

YES

it is impossible to list all the factors, since many are out of your control (ie net congestion at several name switches)

in the mean time my post above is a simply guide line, use it and then adjust  to your DS specs

(Edited by RN Malboeuf at 5:49 pm on Dec. 22, 2001)

(Edited by RN Malboeuf at 3:20 pm on Dec. 23, 2001)

[MugHug 0]

As I stated, in OUR EXPERIENCE we have found that a good connection is only the start of working out what a certain DS can handle in the number of players.

According to your post your server can handle approx. 30 players and therefore you must have a T1 or similar based on your info. The original post was in connection with aDSL. If anything my connection is more closely matched to his then yours.

Therefore I saw a close relationship between his situation and mine and felt I could maybe help.

Remember our experience is based on real-world testing and monitoring and not just figures. We have also found contradictions in OFP MP, which cannot always be explained.

I could sit here and write a thesis on networking (being an active MCSE and employed as an IT Manager), but the purpose was to keep it simple and offer some helpful tips that can be applied in the REAL WORLD, unless everyone is a networking professional and we all want to go crazy over technical points!!!!.

I am not going to debate over your use of 'Totally False' that you chuck around. Not worth the time. Everything I stated could be confirmed, so no point in wasting my time.

I wrote a general non-technical post based upon OUR experience in the hope it may help someone.

It is not the definitive guide nor can it be applied to all situations. It can only help to give ideas to people or to point them in a certain direction.

Your comment about v1.40 is 'Totally False'. V1.40 has only been out for a few days.

Even if you worked for BIS, it is clear that their real-world testing ability is limited considering certain problems with the various patches.

How much real time have you spent monitoring v1.40s performance in relationship to various connections??

How many setups have you tried it on??.

Have you tried it on aDSL, cable, T1 and T3?

Explain why the FPS can go down in what looks like direct relationship to high pings??, or is the #monitor command lying??

Spurting out facts and figures is fine, but try not to lose focus of the whole point of my post. It was to offer real-world experience and tips based upon a cheap typical aDSL setup run and paid for by just me at home (Similar to the original post).

When I started an OFP DS, I wish I had some simple basic tips to help me get it running smoothly.

Not fancy calculations or stories about how wonderful our T1 connection is or that we can host 30 players.

Too many people out there think they are experts!!!!!!!

MugHug

[RN Malboeuf 12]

I have a T3 at work, T1 at home DSL at grandmas, and have hosted DS and hosted as player from each conection and played from each conection

my favorite past time is watching Ping rates to my server or the ones we use

as I stated your posted myths, they are not true they are false

Also if you were correct we would see ping spikes on "busy" servers like radishs and Frag house where we get 15 players in the lobby with a player conecting/diconecting every 20 seconds

yet there is none

can you explain this?

I want a 5000 word thesis

[MugHug 0]

Mmmmm,

'my favorite past time is watching Ping rates to my server or the ones we use'

That explains your fixation with numbers. Try to get out more.

And if I need to explain anything to you, it shows the gaps in your knowledge and experience!!!

I will tell you one thing, I have worked in the field of computers for the last 20 yrs. I still find there are things I do not know. I still find that because something works once, it may not work exactly the same way again. I still find that paper does not always translate to the real world exactly 100%. Maybe my trouble is that I live in the real world!!!

If networking and the Internet was so cut and dry as you seem to put it, them why do companies spend so much money on IT staff and training??

I would like to wish you a merry Xmas and a Happy New Year, but no doubt being the expert, you will have something to say about that!!!.

Anyway, have a nice Xmas and New Year. Oh, on Xmas day stop checking your 'PING' and spend it with your family. Just do not bore them with you tech talk.

MugHug

(Edited by MugHug at 1:56 am on Dec. 24, 2001)

(Edited by MugHug at 1:59 am on Dec. 24, 2001)

[RN Malboeuf 12]

I am with the family now, gone from home for 2 weeks

Try and stick with the discusion and not flame

you still have not explained why well used servers don't get ping spikes when players join ever time

I did, the force Conect command, and it's rarley used since not many know the command line, ASE has this feature and we tested throughly, it does give every one in the game a lurch, and there is still no evidence what so ever that players Pings spike every time some joins

we all ready know that little info is sent to a joining player, they just get mission info, player names, player info if a player has and text sent

this is 10/100th the info needed to create the ping spikes you talk of

you said it your self

DSL, Networked LAN, even if it 300k upload your LAN can affect you this way through various bottle necks, lower end Router (and hub ), tell me whats your data transfer on the router you have at home? how may computers are running

untill you can prove that these things are indeed happing on a proper DS set up don't spread rumors

(Edited by RN Malboeuf at 3:23 pm on Dec. 23, 2001)

[WisdoM 1]

I'm beginning to think Malbeouf is a 10 yr old kid, who acctually knows nothing in reality...but he just likes to argue with everyone Malbeouf you need to learn to chill/grow up or both. personally you thinking you're right all teh time is quite annoying.

[sergeant rock 3]

"my favorite past time is watching Ping rates to my server or the ones we use" - RN Malboeuf

ROFL You crack me up Mal.....

[Angry Radish 0]

Now now, don't feed the trolls:-)

hmm...you know, this forum hasn't had a god thread yet.....

And as a sort-of-used server, We see lag spikes when folks join as well.