Transcript AP - Educause

Technology for High
Performance WLANs
Serving The Needs Of
Higher Education
Agenda
•
Meru Networks – Our Mission
•
Why 802.11 WLANs require QoS for Voice
and Data
•
Meru’s QoS Architecture
•
Comparing QoS Solutions
•
Converged Network Case Study
•
High Performance b/g Network Co-Existence
•
Location Based Services
2
Our Mission
Design and manufacture 3rd generation WLAN
(Wi-Fi ) solutions for voice & data
High Performance:
With
Meru Air Traffic
Control Technology
QoS
5x Number of Voice Calls
High Density
5x Number of Active Users
Transparent Mobility
0 Loss - Handoff
Zero Configuration
Easy Deployment & Management
Comprehensive Security
3
Multi-layers
Meru Wireless LAN
Infrastructure Products
Floor 2
 Coordinated Access Points
► Over-the-Air QoS
► Contention management
Meru AP
AP
Virtual AP
 Controller
► Centralized appliance for
management and security
► RF Interference Management
► Built-in application
Flow-Detectors e.g.
SIP, H.323, Cisco Skinny,
Spectralink SVP
► Location Services
Floor 1
L2 / L3
Backbone
AP
Data Center
Meru Controller
4
Wireless LAN Evolution
Stand Alone
Multi-site
Pervasive
Number of
Clients and
Coverage
Applications • Email, Web
Products /
Technology
• Stand-alone
Access Points
• Email, Web from
different locations
• Centralized
security and
management
5
• Voice and Data
• Business applications
• Primary connectivity
• Video emerging
• High Density
• Application QoS
• Transparent mobility
• Full network integration
Enterprise WLAN Product Evolution
2000-01
1st
Generation
2002-3
2003-4
2nd Generation
3rd Generation
Meru
Cisco 350
Proxim
Linksys
Basic
Connectivity
Stand-alone APs
Cisco 1200+SWAN
Symbol
Aruba, Trapeze,
Airespace …
Generation 1 +
Central Management
Security
Aggregated AP’s
Central Switch/
Appliance
6
Generation 2 +
RF Intelligence
High Density
QoS
Zero Handoff
Coordinated AP’s
Central Switch/
Appliance
University WLAN Requirements
 High Density
► Lecture halls, classrooms, memorial unions, etc
 High Mobility
► Students, faculty, visitors – constant movement
 Data, Voice, and Video
► Data today
► Voice emerging – soft phones, dual mode cell phones
► Video – lecture content, video presentations, etc.
 Integrated Security
► Student / faculty / guest security profiles
► Integration with network access control
► Location based security
7
WLAN Architectures
“Stand Alone” or
Fat” AP
“Thin” or “Fit” AP
WLAN Switch
/ Controller
LAN Switch
“Intelligent” AP
WLAN Switch
/ Controller
Architecture
Stand alone APs,
Centralized or Distributed
Management
Centralized intelligence
Overlay WLAN network,
distributed intelligence,
centralized control
Benefits
Simple implementation for
a few access points,
Centralized Management
Scaleable, central
management
RF coordination, simple
deployment, scaleable,
central management
Target
Markets
SOHO, Medium & large
enterprise
Medium & large enterprise
Medium & large enterprise
8
Key Requirements WLANs
With Standard WiFi Clients
Air Traffic
Control
Technology
QoS
Priority for Voice Calls, Video
High Density
Transparent Mobility
Capacity for Active Users
Zero Loss Handoff
Easy Deployment &
Management
Security
Take Control of the Converged WLAN
9
Issues for Voice & Data over Wi-Fi
Unpredictable behavior over-the-air
1.

Random allocation of air time
Poor performance
2.

Low user density

Low number of voice calls, call quality

Handoff
3.
Security
4.
Difficult to install
5.
Difficult to manage
10
802.11 Challenges for Voice & Data
The 4 fundamental problems that must be
solved to achieve enterprise level
performance for high density Voice and
Data for Wi-Fi networks:
1. Single Cell Contention
2. Contention Across Cells
3. Jitter and Volatile Bandwidth Allocation
4. Slow Handoff
11
Problem 1: Contention within Single Cell
S
Air = Shared
Medium
R
Baseband + Protocol Overhead
8
(Mbps)
Total Bandwidth at Peak
I
I
I
11
5
Contention
Loss
1
20
3
Number of Active Users
• Multiple clients contend for the same shared medium
• While transmitting sender cannot listen for collisions
• As number of calls goes up, collisions increase
• Collisions cause clients to backoff
• Backoff slows down network
• Requires more than scheduling
12
Problem 2:
Co-Channel Interference
Across Multiple Cells
Mbps
Receive Signal at AP
5.5/11
Mbps
Cell
size
Interference Range Is Much Larger
Than Communications Range
1
Mbps
Collision
Domain
CS
Interference
Domain
Distance
13
Problem 3: Jitter & Bandwidth Allocation
Throughput
1 AP + 20 Clients
Channel Access with Today’s 802.11 AP
12
Channel
10
8
6
4
2
5.36 5.38
5.4
5.42 5.44 5.46 5.48
5.5
5.52 5.54 5.56
Time (Sec)
As number of calls goes up:
• Random channel access by clients causes latency & jitter
•
AP gets less bandwidth (only 1/nth of channel)
•
Erratic, unfair access over short term intervals (completely
starved 2 clients)
14
Problem 4: Slow Handoff Across Cells
• Beacon and Probes to join
available ESSID
BSSID = A
• 802.11 Association and
Authentication process
BSSID = B
• 802.1X Authentication or
any other type of security
authentications (includes
Radius or other AAA
servers)
01:00
100ms – 1 sec between handoff
• IP address assignment
15
Meru Networks - QoS Architecture

Global RF Resource
Knowledge
+
Meru QoS
Algorithms
Application Flow
Detection
Global knowledge of interference and resource
usage at AP’s including knowledge of clients
 Time-based accounting, not bandwidth-based
 Inter-cell Coordination

Deep packet inspection for understanding resource
requirements of Application (e.g. SIP/Codec)

Resource management
+
Admission Control
+
Per-flow Scheduling
+
Control Mechanisms
in 802.11 Standard

Uplink and Downlink accounting of packets /
expected packets
 Reservation-based QoS

Virtual carrier sense for uplink reservation/QoS
 Contention-free periods and contention periods.
16
Meru Networks - Air Traffic Control
Contention
Management
Algorithms
Centralized Control
- Global Policies
- Global Coordination
- Central RF Intelligence
- App Flow Inspection
Contention
Suppression for
QoS Flow
Virtual MAC
for
Zero Handoff
MERU MAC
- Local Governance
- Dynamic QoS Flow
Recognition
- Distributed Rogue
Detection & Mitigation
Performed in Controller
Performed in the Access Point
Voice
Client
17
Comparing Control of the Air
How Meru Delivers Over-the-Air QoS
Meru AP
Other APs
 Access to the Lower
MAC is critical to
provide QoS
 Decisions need to be
Mgmt (Auth/Assoc/Probe)
Beaconing
Meru AP
(with
Meru MAC)
Packet Fragmentation
Scheduling/Queuing
Sychronous
Interface
802.11
Phy/RF
Referencedesign AP
Lower MAC (CSMA/CA)
PHY
Asynchronous
Interface
Between SW
And MAC/PHY
Integrated
MAC/PHY
RF
18
made at microsecond
level based on prior
packet air conditions
 Other AP’s queue
packets asynchronously
requiring decisions to
be made several time
intervals prior to
transmission
Air Traffic Control - The Result
Application Flows with Over-the-Air QoS
Channel Access with Meru AP for QoS Flows
Meru AP
C12
C10
C8
C6
C4
AP
5.36
5.38
5.4
5.42
5.44
5.46
5.48
Time (Sec)
5.5
 Predictable channel access
 Predictable and low jitter
 Support for higher number of clients
19
5.52
5.54
5.56
Meru Air Traffic Control Technology
5x More Users
Active Users Per AP
100+
11
8
(Mbps)
Total Bandwidth at Peak
Peak Aggregate Throughput
Meru AP
Performance
5
Contention
Loss
1
5X
20-25
Today’s AP
Performance
3
Today
20-25
Number of Active Users
20
Meru
Meru Air Traffic Control Technology
Over-The-Air QoS: 5X More Voice Calls
Over-the-air
QoS
AP
Standards-based
Over-the-Air
QoS
Wired
QoS
AP
Wired
QoS
~20-30
Voice
Quality
MOS Score
~5-8
4.0+
Generic Access Point +
Standard Client
Meru AP +
Standard Client
Data and voice typically on
Separate channels/network
Converged Network - voice
and data on same channels
21
Meru Quality of Service - Results
Throughput
1 AP + 20 Clients
Throughput
1 Meru AP + 20 Clients
 Industry leading aggregate throughput at density
 Predictable, uniformly fair throughput across all clients
►
Other AP’s erratic, unfair access over short term intervals
completely starved 2 clients
 4X less loss rate (2% - 2.5%)
►
Versus other AP’s 8% loss rate
22
Meru Air Traffic Control Technology
Results - Zero Loss Handoff
Meru WLAN
Today’s WLAN
Virtual AP Architecture
BSSID = A
BSSID = M
BSSID = B
BSSID = M
00:00
01:00
100ms – 1 sec between handoff
No Handoff For Client
23
Meru Quality of Service - Summary
 Works with all standard 802.11 Wi-Fi clients
 Fine grained upstream and downstream
over-the-air QoS with easy provisioning
►
Voice flow detectors (SIP, H.323, Vocera,
Spectralink, Cisco)
►
►
Application QoS Rules
►
►
Real-time highest priority
Real-time, user-configurable rules
Client Fairness
►
►
8 priority queues
Optimized throughput with Meru Air Traffic Control
algorithms for predictable performance
24
How Meru Over-the-Air QoS
Compares to Others
Meru
802.11e / WMM
Today’s AP’s /
“WLAN Switches”
Global RF Knowledge and
Inter-cell Coordination
Yes
--
--
Application Flow Detection
and Classification
Yes (Dynamic)
--
Static ESSID-based or
Filters
Yes
--
--
Downlink (AP to Client)
Reservation-based
True over-the air QoS
Low-scale
Simple Priority of
packets
Uplink (Client to AP)
Reservation-based
True over-the air QoS
Low-scale
With Client-side HW/SW
--
Admission Control
25
Customer Case Study
Jackson Memorial Hospital
A Meru Customer Success Story
“
Creating a Wireless LAN with
better utilization across different
applications is the right move for
companies today. Enterprises
require third generation Wireless
LAN products with coordinated
Access Points that permit greater
scalability and centralized
management. This will lead to a
reduction in the overall costs of
wireless infrastructure while
improving performance.
Rachna Ahlawat, Principal
Analyst, Gartner Inc.
27
”
The Jackson Memorial Hospital
Wi-Fi Challenge
Key Requirements
 An indoor WLAN
solution that could
reliably co-exist with its
existing outdoor AP’s
 Future-proof system to
support data today and
voice in the future.
 Support for high user
density and broad
range of devices
Why Other Systems Fall Short
 Inability to manage
contention needed to
support high density
environments
 Cannot operate on a single
channel to avoid
interference with outdoor AP
 Unable to deliver over-theair QoS needed for missioncritical applications
28
Single Channel Deployment
Leverage Existing Wired & Wireless Investments
PAC Building
Network Admin Building
Floor 7
Meru AP
Laptops
Floor 6
 Meru system seamlessly
connected to existing
wired Cisco switches
and works with any
standard 802.11 client
(phone, pda, laptops)
 Dynamic over-the-air
QoS supports reliable
data services today and
high-performance voice
and video in the future
Cisco Catalyst
Switch
Tablet PCs
Floor 5
PDAs
Virtual AP
Floor 4
WiFi
Phones
Floor 3
Vocera
Badges
Cisco Catalyst
Switch
Floor 2
Laptops
Data Center
Netscreen Meru
SSL VPN Controller
29
Vocera
PBX
System Server
Outdoor
Cisco Catalyst
6500
Outdoor
AP
High Performance b/g Network
Co-Existence
Why 802.11 b/g Co-Existence?
 Backwards compatibility of b clients
► Large and growing installed base of b
clients (Millions)
 Utilize same AP infrastructure
► No new AP installations
► No RF re-planning
 Higher channel efficiency for g networks
► Leverages the g network speed – 54 Mbps
31
The b/g Co-Existence Problems
 Significant Co-Channel interference
Only 3 spectrally independent channels
► Coverage required for high data rates
►
 802.11b slows down g clients
►
g client throughput reduced by 50%+
32
802.11b Slows g Clients
 b client preamble and header impact control and data
periods for g clients
 Significant reduction in data rate – greater than 50%
802.11b Only
Preamble
PLCP
Data
Preamble PLCP
ACK
CCK
X micro sec.
802.11g Only
Pre
P
L
C
P
Data
Pre
P
L ACK
C
P
OFDM
> 2X micro sec.
g Client
802.11g/b
b Client
Preamble PLCP
CTS
Pre
P
L
C
P
Data
Pre
Virtual Carrier Sense
Carrier Sense
33
P
L ACK
C
P
Concurrent High Performance
 Separate 802.11b and 802.11g networks into
different BSSIDs
►
Logically isolate b and g clients
 Creates packet level interoperation
► Controlled channel access
► g only window
► b only window
 Adaptively determine the window period
► Protocol content
► Flow-level info (upstream & downstream)
► Number of b clients
► Number of g clients
34
Deployment Architecture
Virtual Wireless
Subnet
DHCP
Server
Meru Controller
RADIUS
Server
 Separate BSSIDs
Routed
Core
Meru AP to
controller tunnels
established over
routed
infrastructure
802.11b
ESSID #1
AP
AP
b
b
802.11g
ESSID #2
AP
g
g
g
35
for b and g clients
 AP’s can advertise
each b and g
network (2 BSSIDs)
 APs control channel
access based on
required b and g
resources
 APs utilize adaptive
control algorithms to
determine window
period
Meru Eliminates Trade-Off Between
Backward Compatibility and g Throughput
UDP
g
TCP
1g +1b
g client perf.
TCP
1g
(with b clients present)
~22.6
~10.1
~15.8
~10.3
~14.8
~3.4
Vendor C
Meru
Vendor C
Source: Meru Lab Tests
36
Meru
Vendor C
Meru
Summary
 Breakthrough Air Traffic Control Technology
Delivers Concurrent 802.11b and 802.11g with
High Performance
►
Simplify 802.11g deployment – no new APs
►
Highest 802.11g throughput in mixed b/g networks
►
Leverage deployed 802.11b clients
►
Eliminate user performance compromise
37
Location Based Services
Planning / Site Survey – “Snap Shot”
 Create Network Plan:
► Upload map (.jpg or .png file, no need
for CAD drawings)
► Draw walls and other obstacles
(optional)
► Place access points on the map
► Simulate the network coverage

Perform Site Survey
►
►
►
Upload map (.jpg or .png file, no need for CAD drawings)
Deploy the APs as per plan Survey the site - Measure the coverage
Fix the coverage holes if any by adding APs or adjusting antennas
39
RF Visualization – “Real-Time”
Visualize Coverage
• Visualize coverage based on signal strength, data rate,
• Determine which areas support given ESSID, or channels
• Visualize network performance and coverage holes
40
Location Tracking Applications
 Real-time location of Rogues, clients
► Pinpoint rogue device (AP or client) to specific location (in
a cubicle, in the hallway, outside the building)
► Allow connectivity only when client at specific location (e.g.
inside building)
 Real-time capacity management/troubleshooting
► Identify relevant portion of a network for capacity
adjustment or troubleshooting based on caller’s location
 Mobile asset tracking
► Locate critical equipment or assets in hospital,
manufacturing, retail environments
 E-911 support
► Meet regulatory requirements for calls that require
emergency dispatch
41
Location Tracking Technology
 Traditional approaches:
► Closest AP – find the AP that hears a signal the
loudest
►
►
Triangulation – overlap coverage from 3 different APs
►
►
►
Very coarse granularity (point in 60’x60’ or 3600 sq ft area)
Granularity of ~ 30’
Challenges: Reflection, attenuation, multi-path
RF-Fingerprinting – predict signal strength at every
grid point, and match against it
►
►
Hours of RF signature training ( ‘can you hear me now?’
approach)
Granularity of ~10’
42
Summary
 Over-the Air QoS is required for Converged
WLAN networks
 Breakthrough Technology Delivers Concurrent
802.11b and 802.11g with High Performance
 “Stay tuned” for Location based WLAN services
43
Thank You
Nate Walker
Director, Product Management
[email protected]