Apricot2001 Effectiveness of VLAN Chan Wai Kok (wkchan@mmu.edu.my) Faculty of Information Technology Salim Beg (salim@mmu.edu.my) Faculty of Engineering Multimedia University - PowerPoint PPT Presentation

About This Presentation
Title:

Apricot2001 Effectiveness of VLAN Chan Wai Kok (wkchan@mmu.edu.my) Faculty of Information Technology Salim Beg (salim@mmu.edu.my) Faculty of Engineering Multimedia University

Description:

Simplify the process of add, moves and changes ... Term as IGMP spoofing. Dynamic filtering of IP Multicast group at switch port level. ... – PowerPoint PPT presentation

Number of Views:36
Avg rating:3.0/5.0
Slides: 26
Provided by: scotth61
Category:

less

Transcript and Presenter's Notes

Title: Apricot2001 Effectiveness of VLAN Chan Wai Kok (wkchan@mmu.edu.my) Faculty of Information Technology Salim Beg (salim@mmu.edu.my) Faculty of Engineering Multimedia University


1
Apricot2001 Effectiveness of VLAN Chan Wai
Kok (wkchan_at_mmu.edu.my)Faculty of Information
TechnologySalim Beg (salim_at_mmu.edu.my) Faculty
of EngineeringMultimedia University
2
The benefit of VLAN
  • Simplify the process of add, moves and changes
  • Provide broadcast, multicast traffic isolation by
    grouping similar traffic types to a VLAN.
  • Provide some form of security

3
Goals of this presentation
  • To find out the broadcast traffic trend
  • To study the effectiveness of VLAN in broadcast
    isolation

4
Why Broadcast isolation is important ?
  • On a 1 Gbps capacity backbone network
  • If 1 traffic is (broadcast multicast unknown
    unicast), all end stations with 10Mbps link may
    be congested (but backbone have more capacity!!!)
  • Broadcast frames will take up CPU resources. E.g
    On Pentium 120Mhz,
  • 1000 broadcast pps - 15 CPU resource
  • 3000 broadcast pps - 28 CPU resource

5
Related Work(Will E. Leland Daniel V Wilson
Bellcore 1989 - 1990)
  • collected LAN traffic data for long time scale
    and analyzed the trend
  • Their showed that there is a great disparity in
    the values of peak to mean ratio for packet
    arrival rate. (73 to 861 for 5 sec to 5 milisec
    respectively)
  • LAN traffic is Bursty
  • Focus on behavior of total Ethernet traffic

6
LAN Traffic (hourly)
7
LAN Traffic (busiest hour)
8
LAN Traffic (by minutes)
9
LAN Traffic (Peak to mean ratio)
10
Analysis Method
  • Collect network traffic from few different IP
    Subnet and check the broadcast plus multicast
    traffic quantity.
  • Run an IP Multicast application
  • See how its effect the broadcast plus multicast
    traffic in the IP subnet.
  • Make some conclusions on the result

11
Network Setup
  • A Pentium PC is used as a network analyzer
  • Traces are taken for a few days
  • Network consists of WinNT servers and Win95 PCs.
  • Each IP subnet consists of 60, 85 and 45 hosts
    respectively. 3 different labs.
  • Two Labs are is fully used during office hour.
  • Students accessing WinNT server to work
  • No Novell Server (future is IP)

12
Type of Packets Detected
  • ARP request
  • Spanning Tree Frame (hello frame)
  • Netbios
  • SAP
  • IP Multicast Solicitation

13
Traffic statistic
  • Average bandwidth (extremely low)
  • 1413, 1761, 1381 bps (office hour)
  • 809, 789, 406 bps (non office hour)
  • Maximum bandwidth (1 of 10Mbps)
  • 31920, 125408, 130752 bps (office hour)
  • 29824, 70640, 38608 bps (non office hour)
  • Majority of the traffic is background traffic
  • Generally, broadcast traffic is very low

14
Traffic Statistic (no IP Multicast)
15
Traffic Statistic (no IP Multicast)
16
Traffic Statistic (no IP Multicast)
17
Traffic Statistic with IP Multicast application
  • One IP Multicast Channel sending audio and text.
  • Average bandwidth - 112 Kbps
  • Maximum Bandwidth - 155 Kbps
  • Dramatic increased of (broadcast plus multicast)
    bandwidth when IP Multicast is ran.
  • Traffic isolation will required IP Multicast VLAN

18
Traffic Statistic with IP Multicast application
19
IP Multicast VLAN
  • Router will duplicate IP Multicast packet to all
    IP Subnet that joined in IP Multicast group
  • Some switch ports have multiple IP Subnet VLAN
    memberships. Therefore, may receive in duplicated
    IP Multicast packet
  • IP Multicast VLAN is patented by 3com US Patent
    No 5,818,838
  • Term as IGMP spoofing
  • Dynamic filtering of IP Multicast group at switch
    port level.
  • Available in many commercial switches.

20
IP Multicast Application
  • Who is running IP Multicast at the 1st place ?
  • A 4.5 year studies on MBone traffic shows
  • 90 of users joined multicast group less than 74
    times
  • top 1 joined MBone group gt 350 times
  • 20 IP addresses - at least 1 hour per connection
  • 70 IP addresses - average 1 min per connection

21
IP Multicast Application
  • Over 4.5 year
  • 20 of users joined MBone more than 1 day
  • 5 of users joined gt 300 hours
  • A total of 33,545 unique IP is seen in 4.5 year
  • Jan 1998 - 10,600 unique IP seen
  • ( gt 20,000 IP didnt joined any multicast session
    for more than 1 year)
  • Jan 1999 - 5,000 unique IP seen

22
Conclusion
  • When no IP Multicast application
  • Total Broadcast and multicast traffic is very low
  • problem of broadcast isolation does not arise at
    the first place
  • VLAN is of little use

23
Conclusion
  • When IP Multicast is deployed
  • Dramatic increase in mean bandwidth
  • VLAN is useful only when IP Multicast application
    is deployed
  • Make sure that current/future switch support IP
    Multicast VLAN
  • However, IP Multicast application is still not
    popular yet.
  • Studies have shown MBone is growing slowly and it
    uses is not widespread.

24
References
  • http//imj.ucsb.edu/publications.html
  • BGPK. Almeroth, "A Long-Term Analysis of Growth
    and Usage Patterns in the Multicast Backbone
    (MBone)", IEEE INFOCOM '00, Tel Aviv, ISRAEL,
    March 2000
  • http//www.patents.ibm.com
  • http//www.argreenhouse.com/papers/wel/
  • Will E. Leland, Daniel V. Wilson, High
    Time-resolution Measurement and Analysis of LAN
    Traffic Implications for LAN Interconnection,
    Proceedings IEEE Infocom 1991, Bal Harbour,
    Florida, April 1991, pp.1360-1366. building and
    forwarding

25
Question ??
Question can be forwarded to wkchan_at_mmu.edu.my
Write a Comment
User Comments (0)
About PowerShow.com