The technological limits of online gaming ...

[Gannon]

I am curious as to just how limited online gaming is at present. You see MMO's that shy away from action oriented gaming towards the client server and "automatic" computer controlled characters that automatically perform things due to latency, but then you have games like planetside that try to bring "real time" action gaming to MMO's.

Just what are the limits here, is latency really the only limiting factor? If we all had fibre optic connections would there still be lag? I understand that lag is the result of the amount of data one must recieve per unit of time to accurately describe what the objects are doing and the states of those objects, basically you're getting updates to the local objects on your computer.

But even if we all had FTTO would you still have low player counts and whatnot is it possible to accurately have each objects position and state represented so accurately that any latency is negligable?

I always wondered what games like Quake3 or UT2004 would look like if you overlayed the data of everyones machines at the same time and overlay transparent models of the players as "ghosts" to see there true positions on their machines at Time X.

How long and what kind of hardware would it take to get negligable delays/object update times so that when you aim and fire or perform an action it's so on the verge of being real time its negligable?

 

[CycloWizard]

The theoretical minimum latency is just the distance separating each member divided by the speed of light. For example, if I want to play someone on the exact opposite side of the planet (~12,500 miles away), my ping would still be 67 ms (12500 mi/18600 mi/sec). That's pretty good, but as any Q3 player can tell you, a ping of <=30 is definitely preferable.

For any real system, you'd have to include the time added to process and retransmit at each node between yourself and the other members. This would include every router, firewall, or what have you. Due to the current structure (or lack thereof) of the internet in the US, this is usually pretty substantial overhead. I'm not sure what the theoretical limit of processing and retransmission would be, but there are enough computer-savvy folks here to give you a real answer.

 

[everman]

Many games have tried to go around the latency limitation ( In MMOS) by queing actions. This tends to work pretty well, but it's not for everyone.
WoW was going to do forced localized servers, but I don't think they went through with it.
And more clients = more power requirement. Some systems on drawing boards are currently too complex to implement, so more storage is needed. These are temporary barriers.

 

[Gannon]

Yeah I wonder if some real scientific breakthroughs will have to occur or occur just because games and entertainment is such an engineering and problem solving enterprise when you start to make games the scale of MMO's.

 

[I Saw OJ]

Originally posted by: CycloWizard
The theoretical minimum latency is just the distance separating each member divided by the speed of light. For example, if I want to play someone on the exact opposite side of the planet (~12,500 miles away), my ping would still be 67 ms (12500 mi/18600 mi/sec). That's pretty good, but as any Q3 player can tell you, a ping of <=30 is definitely preferable.

That would be true i suppose if you had a direct connection between those computers. But the paths those two computers would take to communicate with each other would definatly not be so. So you would also have to take into account how each "player" finds his path to the game server.


seth

 

[uOpt]

The speed of light is actually the problem here. If you do something that requires an answer packet you can do whatever you want you'll have that limit and your round-trip is not neglectable.

I don't think that wormhole transporting information will make it into a best-buy-able router product anytime soon

And think of people who want to play from a base on the moon or on Mars. That would just suck.

 

[byosys]

12500 mi/18600 mi/sec

Your missing a 0 in there. 186,000, not 18,600.

 

[Veramocor]

Originally posted by: CycloWizard
The theoretical minimum latency is just the distance separating each member divided by the speed of light. For example, if I want to play someone on the exact opposite side of the planet (~12,500 miles away), my ping would still be 67 ms (12500 mi/18600 mi/sec). That's pretty good, but as any Q3 player can tell you, a ping of <=30 is definitely preferable.

For any real system, you'd have to include the time added to process and retransmit at each node between yourself and the other members. This would include every router, firewall, or what have you. Due to the current structure (or lack thereof) of the internet in the US, this is usually pretty substantial overhead. I'm not sure what the theoretical limit of processing and retransmission would be, but there are enough computer-savvy folks here to give you a real answer.

Technically the shortest distance would be a direct line through the center of the Earth. C=2*12,500 miles C= pi*D. D = 25,000. Using your formula (7976 mi/186,000mi/sec), or 43 ms.
Of course we can't send waves through the earth and recieve them yet, but the theoretical speed limitation is 43 ms not 67ms. Perhaps a neutrino communication system could be a point to point system. Not technically possible for 100's of years. But it is still physically possible to do. Going faster than 43 ms would be physically impossible to do.

So a direct line communications at 43 ms would not be under your 30 ms Q3 playing threshold.. Arghhh no way to defeat lag.

 

[Sacrifice]

Lag caused by connections is not the only limit in Online gaming, there are other restrictions such as:

- Client (local)
- Server cluster (soft/hardware)
- ISP's on both sides.

An good example i can take is an online game i play on a regular basis: www.eve-online.com, an game that uses an graphical interface and where the data is stored on the servers. Some issue's that EVE online had so far:

- Hardware:

The current cluster exists out of 40 SOL machines in adition to their support servers, some FAStT600 with EXP700 run and some IBM xSeries each configured with 4 Intel Xeon 3.0GHz processors and 16GB of RAM and IBM xSeries 335, they got Cisco 7400 routers. In the cluster they have a capacity of 16TB fiber disks

In their hardware they had capacity issue's, routers which were breaking down at 20kpps and so on.

- ISP eve side:

Some DDOS actions were performed on the server cluster, some nodes in front of it. This while their connected to the UK backbones.

- Routing companies between client and server cluster: Level3.net caused alot of packetloss and from what i heard was that some blades caused the connection to flap, at some times i had 65% packetloss and ping times of 600ms.

ISP client side:

While i know some guys who have direct backbone acces at their computers, of which one even hosts a teamspeak server on 19 Gbps of direct fiber to 11 diffrent Backbone carriers, most people do not have that connection but connections like 300-400Kbps and some even have old 56k modems or ISDN. Not even to mention that my personal experience with ISP's for customers is that there is a poor performance on the network mostly done by cheap materials used (coax, old data centers etc), bad investment, poor knowledge by some people on the NOC teams etc. This can cause lag, dropping nodes and connectivity issues.

Client side:

Specially for ongoing development for games like: Lineage 2, Everquest, UO, Starwars galaxies, WOW, Eve online there are bugs and bad coding. These games are forced by the community, the Sales of the product and many other factors to keep development going and sometimes within a time limit. 3 months is a standard for a large expansion pack, this means coding the product, testing it on a test server, adjusting it, balancing and eventually deploying the patch. I have yet to discover a MMORPG or MMOG that does not have fatal bugs that cause lag or other issues. Hell, even the deployment from a patch causes lag because everyone is going to download the patch which gives a higher loadbalance than normally playing the game, and everyone is offcourse going to download that patch the moment the server came up in the first minute it came up after a long 24 hour wait.

But not only the game client itself, but also other components are to blame for lag and other issue's like BSOD and so on. Hardware, OS, drivers and running background services are the cause of issue's that sometimes truelly amaze me. Like i once had a BSOD due to a certain sound made by a certain gun in a game, which was easy fixed by adjusting the soundcards hardware acceleration.

Most games atleast demand 1.8ghz cpu and 256mb ddr ram minimum, but optimal performance is gained with 2.4ghz and 1gig ddr ram and a nice 128mb/256mb videocard. Some people do not have the money to have a nice 3.2ghz PCI express system with 2gb ddr ram, and run a system with 1 gig and 128mb or even 64mb. Then there are the people with kazaa and its spyware, running background services which require resources, settings, drivers which arent up to date, OS which is messed up etc etc, we all know it i suppose.

So to come back to your question, Just what are the limits here, is latency really the only limiting factor? I hope i answered atleast some questions. Personally i think that an ideal connection, game and servercluster which causes the minimal ammount of lag is something that we currently are close to currently. But it takes time, efford and investment from everyone involved to deliver a product that can run with minimum ammounts of latency.

- sorry for my simplistic terms, but my foreign language. isnt english. -

 

[Googer]

Originally posted by: CycloWizard
The theoretical minimum latency is just the distance separating each member divided by the speed of light. For example, if I want to play someone on the exact opposite side of the planet (~12,500 miles away), my ping would still be 67 ms (12500 mi/18600 mi/sec). That's pretty good, but as any Q3 player can tell you, a ping of <=30 is definitely preferable.

For any real system, you'd have to include the time added to process and retransmit at each node between yourself and the other members. This would include every router, firewall, or what have you. Due to the current structure (or lack thereof) of the internet in the US, this is usually pretty substantial overhead. I'm not sure what the theoretical limit of processing and retransmission would be, but there are enough computer-savvy folks here to give you a real answer.

Good Job.

Some say that the speed of light is faster than that.

Most data that is transmitted to or from the other side of the world is done by high latecy satellites. Just ask any one in the military there is ususly a 1-3 second delay between the guy in the usa and the equipment he is controling oveseas.

 

[Safeway]

Originally posted by: Googer

Originally posted by: CycloWizard
The theoretical minimum latency is just the distance separating each member divided by the speed of light. For example, if I want to play someone on the exact opposite side of the planet (~12,500 miles away), my ping would still be 67 ms (12500 mi/18600 mi/sec). That's pretty good, but as any Q3 player can tell you, a ping of <=30 is definitely preferable.

For any real system, you'd have to include the time added to process and retransmit at each node between yourself and the other members. This would include every router, firewall, or what have you. Due to the current structure (or lack thereof) of the internet in the US, this is usually pretty substantial overhead. I'm not sure what the theoretical limit of processing and retransmission would be, but there are enough computer-savvy folks here to give you a real answer.

Good Job.

Some say that the speed of light is faster than that.

Most data that is transmitted to or from the other side of the world is done by high latecy satellites. Just ask any one in the military there is ususly a 1-3 second delay between the guy in the usa and the equipment he is controling oveseas.

"Some say the speed of light is faster" The fscking speed of light is measured, known, constant (in vacuum).

 

[Googer]

Originally posted by: Safeway

Originally posted by: Googer

Originally posted by: CycloWizard
The theoretical minimum latency is just the distance separating each member divided by the speed of light. For example, if I want to play someone on the exact opposite side of the planet (~12,500 miles away), my ping would still be 67 ms (12500 mi/18600 mi/sec). That's pretty good, but as any Q3 player can tell you, a ping of <=30 is definitely preferable.

For any real system, you'd have to include the time added to process and retransmit at each node between yourself and the other members. This would include every router, firewall, or what have you. Due to the current structure (or lack thereof) of the internet in the US, this is usually pretty substantial overhead. I'm not sure what the theoretical limit of processing and retransmission would be, but there are enough computer-savvy folks here to give you a real answer.

Good Job.

"Some say that the speed of light is faster than that".
Most data that is transmitted to or from the other side of the world is done by high latecy satellites. Just ask any one in the military there is ususly a 1-3 second delay between the guy in the usa and the equipment he is controling oveseas.

"Some say the speed of light is faster" The fscking speed of light is measured, known, constant (in vacuum).

I guess that you have not been following up on the latest scientifc breakthroughs, scientists have been able to measure and change the speed at which light travels. They have been able in a lab to slow it down to the point where they can almost see it move across the room. If I remember correctly.

"Some say that the speed of light is faster than that".
I WAS REMARKING ON HIS ERROR, (12500 mi/18600 mi/sec) IF YOU DID NOT GET THE JOKE.

 

[bobsmith1492]

The speed of light in a vacuum IS a constant. They were able to 'slow' its speed by transmitting it through a material that absorbs and reemits light very slowly so that the light is apparently traveling more slowly. Light always travels at the same speed; it is only by being absorbed by atoms and then being reemitted that it appears to travel more slowly.