IT@Intel Brief

Intel Information Technology

Wireless LAN as the Primary
WLAN Design
Network
June 2006
Intel IT’s vision for a truly mobile workforce inspired a groundbreaking technology transition to
a unified network architecture in which wireless becomes the primary network access method
for data, voice, and video.

We are deploying this primary wireless Profile: Primary Wireless

infrastructure at one of our large 5,000+ • Integrated architecture for data,
user campuses. Early results show that the voice, and video
wireless LAN (WLAN) is capable of supporting
• Application performance comparable
all our planned business applications. Users
find performance comparable to the wired
to wired LAN
LAN when using common office applications • Deployment across 5,000+ user
and actually prefer using the WLAN overall. campus
We expect substantial cost savings due to
• Expected savings of up to 21%
a simplified infrastructure as we move to a
on new infrastructure
more intelligently managed architecture.

Intel Environment
Primary Wireless LAN

WiFi Phones Soft Phones

802.11a and Headsets

Smart Phones
WiFi/Cellular
Laptops
802.11a/g

WAPs
Cellular Tower 802.11a/g

WLAN Controllers
Smart Phones Laptop
WiFi/Cellular

PBX

Data
Internet
Center
Wireless Access Point (WAP)
802.11a/g IP PBX IP Phones

Home Office or Hot Spot

Figure 1. Our primary wireless architecture.

IT@Intel
Wireless Strategy
Establishing wireless as the primary network
access method for business software applications,
voice over IP (VoIP), and multicast video within
one of our large campuses is a major change in
direction for Intel’s network architecture.
Until recently, we deployed wireless LANs based
on 10-Mbps 802.11b, which are in high demand
among our 70,000 mobile users. However,
because they are slower-performing than wired
LANs, our employees typically use them as a
secondary access method. We also use a virtual
private network (VPN) for WLAN security, which
further hinders performance and increases
operational costs. Additionally, having to maintain
separate LAN, WLAN, and phone infrastructures
is becoming increasingly expensive.
To provide an infrastructure capable of
supporting primary wireless, we are designing
a new architecture that integrates wireless and
wired LANs (see Figure 1), providing wireless
performance comparable to the wired LAN,
eliminating excess infrastructure, and reducing
cost. We plan to greatly reduce the number of
wired LAN and phone ports, replacing them with
802.11a wireless connections.
We are increasing the density of access points
to improve coverage and extend network
capacity and performance. The 802.11a standard
provides a nominal 54 Mbps of bandwidth, which
decreases with distance from the access point.
We have designed the network to provide a
minimum of 36 Mbps per access point, shared
between 12 to 22 simultaneous connections,
>10 seconds
<10 seconds
Network Roundtrip Time (RTT)
1 second
100 ms
1 ms
Difference in WLAN vs. LAN
232 µs
0 ms
Conversation
Figure 2. Network performance compared with industry expectations. Wireless and
wired LAN response times are well under expected industry response times for various types
of applications.
with a goal of providing each user at least
5 Mbps bandwidth 90 percent of the time.
We are using technology based on emerging
standards to make network access faster and
easier. We have applied IEEE WiFi standards
(such as 802.11i and 802.1x) to secure and
automatically connect users, eliminating the
need for a VPN. We are using 802.11e Quality of
Service (QoS) to deliver a better voice experience
to the user and secure roaming for wireless VoIP
to ensure that users have uninterrupted service
as they move about the campus.
Current Development
Over the past 18 months, we developed and
installed a campus-wide design spanning five
buildings housing more than 5,000 users. The
self-healing network provides campus-wide
resilience, automatically restoring service in the
event of a failure in part of the infrastructure.
We are supporting users across the campus,
supplying them with 802.11a-enabled notebook
computers based on Intel® Centrino® mobile
technology. This will provide us with a large user
base to validate and fine-tune our architecture.
To evaluate the network design, we conducted
studies of network utilization and performance
characterization of typical office applications,
such as e-mail and calendaring, file and print
services, and Web access.
Results
During normal use of office applications, user
bandwidth consumption was considerably below
Industry Expected RTTS
RTT in WLAN
RTT in LAN
Interactive Streaming Background/
Best Effort
the overall capacity of the WLAN, indicating
that the network is capable of supporting these
and other planned applications. From a user
perspective, performance of the WLAN was
indistinguishable from the wired LAN, and overall,
users preferred the WLAN to the wired network.
Users’ perception of performance is closely related
to the type of application they are using, and this
is reflected in the differing industry expectations
of acceptable network delay for voice, video, and
typical business applications. Although the primary
wireless network was slower than the wired LAN,
the difference was below the threshold at which
users begin to notice delays for any of these
applications, as shown in Figure 2.
Analysis of network traffic showed that users used
the same mix of typical office applications on the
wireless LAN as on the wired LAN and used them
to a similar extent, as shown in Figure 3.
During use of these applications, user consumption
of wireless bandwidth remained well within our
design goals. Average per-user bandwidth use
varied from about 0.25 Mbps for Web access to
1.1 Mbps for real-time collaborative applications.
In all cases, bandwidth consumption was much
less than the available capacity, confirming that
primary wireless is capable of delivering the
bandwidth that our users require.
The network performance and other improvements
resulted in high user satisfaction ratings in response

Relative Application Use in One-Month Period

8% 8%
File/Print File/Print

15% 15%
LDAP LDAP
39%
E-mail Service 42%
E-mail Service

20% 18%
Remote Remote
Procedure Calls Procedure Calls

13% 6% 9% 8%
DNS HTTP DNS HTTP
<1% Database Access
<1 % Database Access
Wireless By Event Count Wired By Event Count
Total 1.2 million events Total 650,000 events

33% 45%
File/Print File/Print

3% LDAP
3% LDAP
16%
36% Remote
E-mail Service 10%
Procedure Calls Remote
27%
E-mail Service Procedure Calls

14% <1%
10% HTTP Database Access
HTTP <1%
Database Access
<1% DNS <1% DNS
Wireless By Volume Wired By Volume
Total Tx 55.0 million bytes + Total Rx 66.6 million bytes Total Tx 47.4 million bytes + Total Rx 114.4 million bytes

Figure 3. Use of applications on wireless and wired LANs. We see employees successfully using the same applications on both wired
and wireless LANs. However, we see increased use of the network on wireless LANs as users take their notebooks with them.
to surveys. Users rated the wireless LAN comparable to the wired LAN 6.00
Wired
or better in several key areas, as shown in Figure 4. When employees Wireless

were asked which they preferred, they expressed a preference for the 5.75

wireless network overall, saying it was easier to use. Because of this, they

Satisfaction Rating
generally preferred to use the WLAN rather than the wired LAN wherever 5.50

they could—even, in many cases, when at their desks, as shown in Figure 4.

5.25
Users gave the WLAN high points for ease and speed of network
connection. Our previous WLANs used a VPN, which required an additional 5.00
network logon step and impacted performance; users liked the fact that
the new WLAN connected them more quickly and automatically through 4.75
802.1x authentication. Network Ease and Stability and Network Overall
Availability Speed of Reliability Security Satisfaction
We also received many comments from users that indicated they were Connecting
Satisfaction Vectors
taking advantage of the improved wireless network to change the way
they worked. Faster connection time made a big difference in conference Figure 4. User satisfaction comparing WLAN with wired LAN
room meetings where users wanted to quickly get online to access
information or send e-mail. As a result, users said they were more likely
to take their notebooks with them and connect from different locations Projected Cost Avoidance for Large Building
throughout the day.
$3.5
Current Baseline
Estimated Cost Benefits
As we move to our new architecture throughout the enterprise, we 3.0

expect financial benefits due to a lower cost of network infrastructure.

Currently, we run an average of 2.5 ports to each employee’s workspace: 2.5

Millions of Dollars USD

one LAN and one phone port, with additional ports for redundancy. 100 Mbps LAN Support
Adds, Moves, and Changes
As we move to the new infrastructure, we anticipate progressively
Telephone Maintenance
reducing the number of ports per user from 2.5 ports, to 1 port, to 0.5 2.0
Port Maintenance
port, to 0.25 port per user in new buildings we construct over time. Primary WLAN Hardware
and Direct Expenses
By calculating potential savings in equipment, cabling, maintenance, IT Office WLAN Equipment
1.5
and other infrastructure costs, we estimate saving between 9 and 21 Active Network Equipment
Telephone Equipment
percent when deploying network infrastructure in new buildings, with (TDM-based or VoIP-PBX)
greater savings as building size increases. Figure 5 shows projected costs 1.0 IT Cabling
for a large building.

Future Plans 0.5

Our results to date show that our new primary wireless network
architecture can support the planned applications at the level of $0
performance that our users require. Our users value the wireless Baseline 1 Port 0.50 Port 0.25 Port
network and enhanced mobility, and we expect to start transitioning 2.5 Ports

the enterprise to the new architecture over the next year. Figure 5. Estimated infrastructure cost avoidance due to
As we begin to widely support other applications such as VoIP on primary wireless. We expect infrastructure costs to decrease as
handsets and dual mode devices, we will build in enhancements such as
we build new buildings with progressively increased primary wireless
capability.
a more standards-based roaming, security, and QoS profile. We will describe
these in future papers as our architecture evolves and is deployed.

This paper is for informational purposes only. THIS DOCUMENT IS
PROVIDED "AS IS" WITH NO WARRANTIES WHATSOEVER, INCLUDING
ANY WARRANTY OF MERCHANTABILITY, NONINFRINGEMENT,
FITNESS FOR ANY PARTICULAR PURPOSE, OR ANY WARRANTY
OTHERWISE ARISING OUT OF ANY PROPOSAL, SPECIFICATION OR
SAMPLE. Intel disclaims all liability, including liability for infringement
of any proprietary rights, relating to use of information in this
specification. No license, express or implied, by estoppel or otherwise,
to any intellectual property rights is granted herein.
Intel, the Intel logo, Intel. Leap ahead. and Intel. Leap ahead. logo, and
Centrino are trademarks or registered trademarks of Intel Corporation or
its subsidiaries in the United States and other countries.
*Other names and brands may be claimed as the property of others.
Copyright 2006, Intel Corporation. All rights reserved.
Printed in USA Please Recycle
0606/ARM/RDA/PDF Order Number: 313098-001US