Video Game Traffic Profile

1
Video Game Traffic Profile

• Glenn Harper
• Joseph Sanchez
• Everett Espinoza
• Anthony Baros

2
Video Gaming A Burgeoning Industry

• As of 2002, Video Gaming was a 30 Billion per
  year industry.
• Game sales rival CD, DVD, and Box Office sales.
• Example Sony Playstation 2
• 25 million consoles sold, 2.6 million connected
  (with separate adapter).

3
Online Gaming The Next Big Thing

• SCEA study has noted that the peak time for
  online play is between 5 and 11 pm
• Nielsen (Television ratings) has noted a drop in
  the number of viewers among males 18-24 during
  the same time period.
• The rise of online gaming and video gaming has
  had a huge impact on entertainment landscape.

4
Previous Work

• Many previous studies have been done to measure
  Web or Multimedia traffic
• Game traffic should be different Low latency,
  small packet size, bursty traffic

5
A Challenge to the Existing Infrastructure of the
Internet

• Large amount of users requires a relatively high
  bandwidth, massively scalable method of
  connecting to servers.
• How do gaming developers and system
  administrators meet this need?

6
Profiling Traffic from Online Gaming

• Three focuses for our project
• What type of traffic is used to connect between
  clients and servers?
• How many gamers are connecting, and how does the
  number change throughout the day?
• What is the geographic distribution of online
  gamers?

7
Possible benefits of this study

• Good data will allow the design of networks to
  better accommodate network gaming traffic
  turbulence
• Provide data for more realistic network
  simulations

8
Type of Traffic

• Hypothesis mostly UDP, with TCP connection to
  tell server when clients join and disconnect.
• Tested using network connection monitoring tool
  while observing an online game.

9
Number of Gamers

• Hypothesis Absolute number unknown, but peak
  should be between 5 and 11 pm, as per the SCEA
  study.
• Test by looking at the number of gamers on
  closest servers at various times during the day.

10
Geographic Distribution

• Hypothesis Should see a normal distribution on a
  map of the Southwestern USA.
• Means should be at cities (Albuquerque, Phoenix,
  San Antonio, Denver, etc.)
• Test by taking IPs of servers with low lag, then
  plug those IPs into a location finder.

11
Test Procedures and Results
12
Initially set up server, but no one connected.

• Initially set up server, but no one connected.
• Took different approach
• Looked at servers with latency lt 50, and servers
  located in southwest
• Assumed these would be servers connecting to us.

13
STEAM Screenshot
14
Steam Screenshot with Server
15
Testing results for span of two weeks for

• Number of Players
• Number of Active servers
• Types of network connections occuring during
  gaming
• Locations of various servers

16
(No Transcript)
17
(No Transcript)
18
Results for individual days of the week

• Number of players
• Number of active servers

19
(No Transcript)
20
(No Transcript)
21
(No Transcript)
22
(No Transcript)
23
(No Transcript)
24
(No Transcript)
25
(No Transcript)
26
Mechanism for determining types of gaming
connections

• Used Netmon 1.1, which records the following
• Remote Address
• Local address
• Port number used
• Type of protocol used

27
NETMON Screenshot
28
This software allowed us to determine the various
types of connections utilized throughout game play
29
Sample of data file created by Netmon
30
Types of connections determined

• There are TCP Connections to central Steam
  server. These connections perform basic control
  functions
• Majority connection types are UDP, which contain
  the bulk of the data. UDP connection is the
  basic server-client connection.

31
Geographic Distribution

• http//ws.arin.net/ was used to find out the
  organization which a IP address belonged to and
  thus the location of the servers city

32
Sample Data
OrgName Game Daemons, Inc.
OrgID GAMED
Address P.O. Box 702
City Midway
StateProv UT
PostalCode 84909
Country US
NetRange 69.12.8.0 - 69.12.9.255
 
 
OrgName Cyberbasin Internet Services
OrgID CIS-218
Address 407 N Big Spring
City Midland
StateProv TX
PostalCode 79702
Country US NetRange 65.164.136.0 -
65.164.139.255
33
After looking up the location of 50 IP addresses,
this is the distribution
34
Conclusions

• Network games different than typical Internet
  applications
• Very Interactive
• Traffic flow profile consists mainly of small
  bursts of UDP packets
• For Counter-strike game, we see burstier data
  rate depending upon action
• Packet sizes for server depend upon action