Backgrounder for Policy Discussions on Wireless

1
Backgrounder for Policy Discussions on Wireless

• Terry Gray
• Director, Networks Distributed Computing
• Scott Mah
• Director, Communication Technologies
• February 2004

2
Outline

• Generalities
• Technology Issues
• Policy Issues
• Funding Issues
• Bandwidth Issues

3
Wireless is...

• Addictive (users love it)
• Seductive (appears to be cheaper/easier than it
  is)
• Expensive to scale to an enterprise-class
  solution
• Encouraging enclaves, balkanization
• Rapidly changing technology
• Hard to control
• Hard to secure
• Either parasitic upon, or synergistic with,
  overall campus network infrastructure
• Best seen as needing to parallel history of
  deployment of Internet at the UW
• Becoming mission-critical

4
Key Issues

• Central vs Departmental wifi coexistence
• Technical standards
• Unauthorized access points
• Security policies (protecting others)
• Access control policies (who can use?)
• Funding and accounting policies
• Rented space, student-owned equipment

5
Technology Issues

• Standards
• IEEE 802.11a, b, e, f, g, h, i (and more!)
• IEEE 802.1x, LEAP, PEAP, TLS, TTLS
• Monitoring, management
• RF propagation, interference, pwr mgt
• Security, access control
• Performance, QoS
• Availability, Reliability
• Convergence

6
Agenda for 1/2003 IEEE meeting
7
Impact of VOIP over Wireless

• Separate backbone?
• Campus-wide roaming?
• Quality/Reliability expectations?

8
Policy Issues

• Access control
• Departmental/private nodes
• Who, if not CC under U-TAC policy direction,
  owns/controls RF spectrum?
• Who defines standards and minimum security and
  coexistence policies?
• Who enforces standards minimum security and
  coexistence policies?
• How will an extensible, scalable and sustainable
  model be established

9
Central vs. Departmental Tensions

• CC not out front (wed say not able to be -)
• Inconsistent access policies (private enclaves)
• Inconsistent or non-existent security provisions
• Inconsistent or incompatible technology
• Inconsistent upgrade maintenance policies
• 24-7 management
• Integration with central network infrastructure
• Integration with central authentication
  infrastructure
• Risks to central net infrastructure and nearby
  hosts

10
Private Wireless Nodes on the Campus Net

• Rationale
• Central service not available
• Central wireless service too expensive (can plug
  cheap wireless access point into campus net)
• Central service sometimes more inconvenient for
  visitors
• Central service is an attractive nuisance
• Very special research requirements
• Special security requirements

11
Funding Issues

• Central, departmental, subscription (voluntary or
  mandatory), STF...
• One-time Capital always easier to find than
  operating
• Recharge strategies incent rogue systems
• Dealing with rogue access points dramatically
  increases operational costs and security
  dangers/costs
• Department STF deployments drive costs they
  dont pay (coping and cleanup is an unfunded
  mandate)

12
Cost Factors

• Degree of convergence
• wired and/vs. wifi data vs. wifi telephony
• Security access control
• Technology immaturity, churn
• Management accounting features (exact parallel
  to routers and e-net switches etc, but harder!)
• User support
• Scaling ( and - economies of scale)
• Sustainability

13
Essential Capital Cost Elements

• Physical facilities (e.g. power, cooling,
  pathways, equipment space and antenna space)
• Wireless Access Points (WAPs)
• Dedicated subnets for wireless (wired Ethernets
  to WAPs, switches, routers, security boxes, etc.)
• Access point management system
• Authentication system
• Authentication management system

14
Operational Cost Elements

• UW Staff
• Design
• HW Installation and SW Configuration/updating
• Monitoring and reporting
• Troubleshooting
• Security incident handling (harder w/wireless)
• User Support
• Sustaining underlying wired net. infrastructure
• Vendor
• Maintenance Upgrades (firmware, SW and HW)

15
Case Study MGH (a new and very well wired
facility)

• Size 99,000 ASF
• Classrooms 27 12
• Floors 4
• Access Points 36
• Initial Cost 94,000
• Initial Cost per Classroom 2,500

16
Bandwidth Consequences

• Wireless implies many more computers, PDAs,
  hybrid cell/802.11 devices, etc.
• Steady growth (or maybe even spike, esp. with
  net generation students) in network devices
• Bandwidth needs track
• users
• usage
• apps and objects
• capacity
• Wireless capacity constrains types of apps (for
  now)

17
Performance Comparison from early 2002 Gig
Ethernet can now exceed 900 Mbps
From www.extremetech.com
18
Network Device Growth
Note Most dips reflect lower summer use last
one is a measurement anomaly
19
Network Traffic Growth (linear)
20
Network Traffic Growth (log)
21
Outcomes to Avoid

• Unrealistic security expectations
• Department wireless deployments that...
• Confuse users re who supports what
• Interfere with or destabilize campus network
• Create extra threats to others
• Balkanize net services w/conflicting policies
• Drive U-wide costs no one is underwriting
• Non-scalable or non-sustainable models

22
Questions? / Comments?