Transcript Sales Meeting

Welcome to
Implementing VoIP:
VoIP Network Best Practices
Implementing VoIP:
VoIP Network
Best Practices
Agenda
Introductions
 From Hype to Adoption
 VoIP Myths
 Anatomy of a VoIP Call
 VoIP Metrics
 Lab 1: VoIP in Action
Break
 The Three Phases of Successful VoIP Deployment
 Phase 1: Site Survey and Testing
 Phase 2: Monitoring the Roll Out
 Phase 3: Ongoing Troubleshooting and Maintenance
Break
 Lab 2: VoIP Call Monitoring
 Best Practices Summary
 Network Instruments Solutions Set
 Q&A

© 2006 Network Instruments, LLC
First things first:
Introductions
Douglas Smith





President and Co-Founder of Network Instruments
Oversees
 Finance
 Sales
 Marketing
 Production
Works closely on product design with company
CEO, Roman Oliynyk
A part of the networking community since 1985
Awarded degrees in Math and Economics
from University of Wisconsin-Madison
© 2006 Network Instruments, LLC
Charles Thompson

Manager of Sales Engineering
 Works directly with the Network Instruments sales
team and partner channel to provide…
 Technical expertise
 Professional services
 In-depth product information for enterprise
accounts
 Travels throughout North America conducting
workshops and presentations on network analysis
 Personally trained thousands of network
managers on the Observer product line
© 2006 Network Instruments, LLC
Network Instruments

Founded in 1994

40,000 licenses sold
 ~4K new customers
annually

12 offices worldwide

Sold in over 75 countries

2005 growth
 Overall: 24%
Network Instruments
Company Growth
1994
1995
1996
1997
1998
Celebrating 11 years of continued company growth
© 2006 Network Instruments, LLC
1999
2000
2001
2002
2003
2004
2005
History of Product Innovation

1st affordable Windows-based analyzer

1st distributed software-based protocol analyzer

1st to include SNMP for switched environments

1st 802.11 a/b/g wireless analyzer and remote probe

1st combined wired and wireless solution together

1st to support 64-bit Windows

1st to develop multi-session, multi-user probes

1st to integrate application analysis

1st to develop enterprise-ready VoIP Expert
© 2006 Network Instruments, LLC
Distributed
Architecture
Advantages
Distributed Network Analysis Architecture
© 2006 Network Instruments, LLC
NI-DNA™ - Distributed Network Analysis
Unified Code Set
Two key components: the console,
which displays data and the probe,
which is used for data collection and
processing.
Scalability
Flexibility
Modularity
Affordability
NI-DNA provides Distributed and complete functionality for every type of
Network topology in any location using the most intelligent Analysis tools
© 2006 Network Instruments, LLC
NI-DNA™ - Distributed Network Analysis
Local/Remote Visibility
Observer console includes a
local probe for local analysis
and connects to remote probes.
Visibility
Efficiency
Productivity
Security
NI-DNA provides Distributed and complete functionality for every type of
Network topology in any location using the most intelligent Analysis tools
© 2006 Network Instruments, LLC
NI-DNA™ - Distributed Network Analysis
Multi-Topology Support
Observer’s single user interface
can manage multiple Gigabit
links, 802.11 connections,
10/100/1000 Ethernet and
Wide Area networks.
Adaptability
Simplicity
Transparency
Reliability
NI-DNA provides Distributed and complete functionality for every type of
Network topology in any location using the most intelligent Analysis tools
© 2006 Network Instruments, LLC
Analysis Options
Software
Probe
Top Talkers
MultiHop
Analysis
Connection
Dynamics
GigaStor
10/100/1000
Probe Appliance
VoIP Analysis
WAN and Gigabit
Probe Appliances
Gigabit and WAN
Observer Suite System
© 2006 Network Instruments, LLC
Application Analysis
SNMP Management
Snapshot of Customers
© 2006 Network Instruments, LLC
From Hype to Adoption
From Hype to Adoption
Market researchers expect the number
of VoIP users worldwide to increase
from around five million in 2004
to 200 million subscribers in 2010
http://www.heise.de/english/newsticker/news/64129
© 2006 Network Instruments, LLC
From Hype to Adoption
“Ninety-nine percent of all VoIP
network implementations that fail
do so because IT departments
didn’t do their homework.”
-- William Stofega,
VoIP research director, IDC
© 2006 Network Instruments, LLC
VoIP Monitoring and Analysis Challenges





Current, competing tools were designed for lab use
No method of quickly determining status and health
No mechanism for understanding aggregate call quality
VoIP dependencies are not implemented properly
Separate tools increase learning curve, reduce ROI
Observer 11
Other VoIP tools
© 2006 Network Instruments, LLC
Common VoIP
Myths
VoIP Myths
Myth #1
Running VoIP without
Quality of Service
is acceptable
© 2006 Network Instruments, LLC
Contention = Delay

Why is Quality of Service important?
 Managing VoIP means managing delay

Even a network with large bandwidth
capabilities can have poor call quality due
to network contention

QoS measures can help make VoIP traffic
less susceptible to adverse network conditions

QoS offers VoIP traffic more
consistent availability
© 2006 Network Instruments, LLC
VoIP Myths
Myth #2
No VoIP Site Survey
is necessary
© 2006 Network Instruments, LLC
Site Surveys are Critical

With the decision to implement VoIP, one of two
choices are usually made

IT managers keep network conditions
the same and add VoIP traffic

IT managers upgrade their bandwidth capacity
© 2006 Network Instruments, LLC
Site Surveys are Critical

Result

Does not solve or address potential
VoIP problems

Network adjustments are made after
deployment has begun

Deployment issues can cause
user resistance to VoIP technology

Adding bandwidth may not be necessary
or offer value
© 2006 Network Instruments, LLC
Common VoIP Myths
Myth #3
Voice conversations
are secure
© 2006 Network Instruments, LLC
VoIP Conversations are a Security Risk

VoIP is another piece of network data

Tools are used to capture not only voice conversations
but to also generate audio playback for later use

Higher-end VoIP systems may offer a way to encrypt
data but many existing or lower-end systems do not

VoIP traffic is most vulnerable on the LAN since
Internet WAN traffic is typically through VPNs

Consider wiretapping rules and regulations
© 2006 Network Instruments, LLC
Anatomy of a VoIP Call
How does VoIP work?
H
o
w
V
o
IP
p
h
o
n
e
s
s
e
n
d
a
u
d
io
s
tr
e
a
m
s
o
v
e
S
e
n
d
W
A
N
L
A
N
L
0
1
0
0
1
1
0
1
E
n
c
o
d
ea
n
dp
a
c
k
e
tiz
e
© 2006 Network Instruments, LLC
N
e
tw
o
rk
d
e
la
y
,jitte
r,&
p
a
c
k
e
tlo
s
s
How does VoIP work?

VoIP phones use codecs to translate analog sound
streams into digital packets for transmission

On the receiving end, the codec translates the
packets back to analog

To ensure normal conversations, all of this must
happen as close to real-time as possible
© 2006 Network Instruments, LLC
VoIP Metrics
VoIP Basics

What is VoIP?


Packetized voice traffic sent over an IP network
What challenges does it bring?

Competes with other traffic on the network

A new technology that needs real-time,
consistent monitoring

Sensitive to delay
© 2006 Network Instruments, LLC
VoIP Basics

Understanding VoIP begins with understanding delay

Normal traffic
 Not sensitive to delay
 Example: FTP, HTTP, e-mail, etc.

Tolerant traffic
 Sensitive to delay
 Loss tolerant
 Buffered by receiver
 Example: streaming video, Internet radio, etc.

Real-time traffic
 Delay and loss sensitive
 Example: VoIP
© 2006 Network Instruments, LLC
VoIP Terms

Jitter

R-Factor / MOS

Burstiness / Gap / Gap Duration

QoS / TOS / Precedence

Compression Techniques (Codecs)
© 2006 Network Instruments, LLC
Jitter

What is it?

Jitter is the variation in the time between packets transmitted
and received

For example, if a packet stream leaves a device with 30 ms
packet spacing and arrives with 50 ms packet spacing, the
jitter is 20 ms
Jitterbufferingandpacket loss concealm
ent
IPPhone
N
etw
ork
Jitterbuffer
© 2006 Network Instruments, LLC
8
9
Codec
7 4
Jitter

Why measure it?

Understanding jitter gives hard facts to help
improve call quality

Excessive jitter will confuse callers about who is
speaking and who is listening

Adjusting jitter buffers can help at the
expense of increased latency and thus,
clipping. Jitter buffer overflow will introduce
dropped packets.
© 2006 Network Instruments, LLC
Observer’s Jitter Measurement
In aggregate…
and per call…
© 2006 Network Instruments, LLC
Call Quality Scoring

What is it?

© 2006 Network Instruments, LLC
Industry standard methodologies for
associating a grade to a call
Call Quality Scoring

R-factor
 Identifies live call quality using a single source
of visibility
 Based on E-Model (ITU G.107)
 Scale: 1-100
 Typically the maximum value would be 93.2
after standard degradation




© 2006 Network Instruments, LLC
Codec used
Network delay
Jitter buffer
Packet loss
Call Quality Scoring

MOS
 User satisfaction level with a call
 Takes into account a number of different factors




Scale: 1-5



Handset quality
Ambient noise
Network performance
4.0 and higher considered satisfied
4.5 and higher extremely satisfied
On simulated calls, traffic is captured at the
destination and compared to the original
sent data to identify degradation
© 2006 Network Instruments, LLC
Call Quality Scoring

Why measure Call Quality?
 Provides objective and subjective scores to
evaluate existing conditions to compare with
historical conditions.
© 2006 Network Instruments, LLC
Observer’s Call Quality Scoring
In aggregate…
per call…
and Expert…
© 2006 Network Instruments, LLC
Burstiness and Burst Density

What is it?




A burst is a period of time characterized by high rates of
packet loss
Burst percentage is the % of time bursts are occurring
Burst density is the rate of VoIP data packets lost during a
burst period
Why measure it?


Higher rates affect call quality, especially when coupled
with long average burst duration times
Possible reason for packet loss include network
congestion, media failure, and link failure
© 2006 Network Instruments, LLC
Gap Density and Duration

What is it?





Gaps are the periods between bursts
A gap is a period of time characterized by lower levels of
packet loss than the burst periods that bound it
Gap density is the percent of packet loss during gaps
Average gap duration is measured in time
Why measure it?


Knowing the gap helps define the burst
In most cases, packet loss during gaps is rendered
insignificant by concealment techniques built into the
VoIP infrastructure
© 2006 Network Instruments, LLC
Observer’s Burst and Gap Density
In aggregate…
and per call…
© 2006 Network Instruments, LLC
Settings for QoS / Precedence

Support for multiple definitions of Quality of Service (QoS)
 Also known as Precedence or Type Of Service (TOS)

What is it?
 QoS is a bit setting used by routers and switches to prioritize
packet flow

Why measure it?
 Incorrectly set QoS can
lead to VoIP or other
network contention
 Contention will lead to
delays in packet delivery,
reducing call quality
© 2006 Network Instruments, LLC
Observer’s QoS/TOS/Precedence
In aggregate…
per call…
and Decode…
© 2006 Network Instruments, LLC
Compression Techniques

Codec is a term for Coder/Decoder

Different compression techniques (codecs)
 G.711: 64kbps (no compression)
 G.729: 8kbps
 G.723: 6.3kbps, 5.3kbps

Higher compression reduces R-Factor and MOS but also
reduces potential contention
© 2006 Network Instruments, LLC
Which Codecs Are Used?
In aggregate…
per call…
and Decode…
© 2006 Network Instruments, LLC
How does VoIP work?
H
o
w
V
o
IP
p
h
o
n
e
s
s
e
n
d
a
u
d
io
s
tr
e
a
m
s
o
v
e
S
e
n
d
W
A
N
L
A
N
L
0
1
0
0
1
1
0
1
E
n
c
o
d
ea
n
dp
a
c
k
e
tiz
e
© 2006 Network Instruments, LLC
N
e
tw
o
rk
d
e
la
y
,jitte
r,&
p
a
c
k
e
tlo
s
s
Lab 1
Capture and Decode
of a VoIP Call
Break
Agenda
Introductions
 From Hype to Adoption
 VoIP Myths
 Anatomy of a VoIP Call
 VoIP Metrics
 Lab 1: VoIP in Action
Break
 The Three Phases of Successful VoIP Deployment
 Phase 1: Site Survey and Testing
 Phase 2: Monitoring the Roll Out
 Phase 3: Ongoing Troubleshooting and Maintenance
Break
 Lab 2: VoIP Call Monitoring
 Best Practices Summary
 Network Instruments Solutions Set
 Q&A

© 2006 Network Instruments, LLC
Phase 1:
Site Survey and Testing
Site Surveys are Critical

Conduct a Site Survey to review…
 WAN link bandwidth levels
 Current traffic flows
 View existing switches for bottlenecks and
choke points
 Determine needs through testing and modeling
 Placement of analysis tools

The more you know about your network the better
prepared you are to properly integrate VoIP
© 2006 Network Instruments, LLC
WAN Link Bandwidth Levels
Summary
Port 1 Source DCE
Port 1 Source DTE
Port 2 Source DCE
Port 2 Source DTE
© 2006 Network Instruments, LLC
Long-Term Trending
© 2006 Network Instruments, LLC
Estimate VoIP Impact

How will VoIP traffic affect the network?

First, determine number of potential users
 Assume users spend 20% of their day on the phone
 Includes active calls as well as VM retrieval
 Video will add to utilization
 Varies based on site

Determine the codec in use; for example:

G.711: 64kbps (no compression)
 G.729: 8kbps
 G.723: 6.3kbps, 5.3kbps
© 2006 Network Instruments, LLC
Examples: VoIP Impact

100 users on site, 20% usage = 20 concurrent sessions
 G.711 Codec: 1.28 Mbps
 G.729 Codec: 160 kbps
 G.723 Codec: ~120 kbps

Bandwidth impact using a T1 at 1.54 Mbps
 G.711 Codec: 83%
 G.729 Codec: 10%
 G.723 Codec: 8%
© 2006 Network Instruments, LLC
Current Traffic Flows


Assuming one drop per user, evaluate connection
speeds and current usage
If multiple network drops per user, not applicable
© 2006 Network Instruments, LLC
Find Bottlenecks and Choke Points


Determining switch, router, and device utilization
Review uplinks and shared pipes
© 2006 Network Instruments, LLC
Testing and Modeling

Do a pilot test to generate sample calls in various
network conditions

Capture live data and model hypothetical situations

Switch codecs to find optimal performance

Use Observer’s “What-If” Analysis to predict
response

Here is an example using Observer…
© 2006 Network Instruments, LLC
Testing and Modeling

Review Observer’s UDP Events to find live calls
© 2006 Network Instruments, LLC
Testing and Modeling
G.711
1 User
© 2006 Network Instruments, LLC
Testing and Modeling
G.711
100 Simultaneous
Users
© 2006 Network Instruments, LLC
Testing and Modeling
G.711
1000 Simultaneous
Users
© 2006 Network Instruments, LLC
Placement of Analysis Tools

Where should you place your analyzer tools for
maximum visibility?

Depends on what you’re wanting to see
 Each call includes both client and server
communications

If you need access to all local conversations…
 Use a SPAN session on the access layer
 Assign all VoIP traffic to a dedicated VLAN
© 2006 Network Instruments, LLC
Points of Visibility

Consider a sample network
V
o
IPca
l
m
a
n
a
g
e
r
A
cce
ssla
ye
r
W
e
stC
o
a
st
O
ffice
C
o
resw
itch
M
P
L
S
M
esh
C
o
resw
itch
© 2006 Network Instruments, LLC
Points of Visibility

Capturing local IP traffic shows

Phone’s communication with its local call
manager

Both sides of the full-duplex connection
between local phones

Both sides of the full-duplex connection
between phones located across a WAN
© 2006 Network Instruments, LLC
Points of Visibility
W
estC
oast
Office
VoIPC
all
M
anager
MPLS
A
ccessswitch
Mesh
© 2006 Network Instruments, LLC
Points of Visibility

Need a more coherent view of calls across WAN links?

Use a SPAN session to mirror…


Both the uplink traffic between the core and
MPLS mesh
All traffic flowing to and from the call manager
© 2006 Network Instruments, LLC
Points of Visibility
W
estC
oast
Office
VoIPC
all
M
anager
A
ccessswitch
MPLS
Mesh Core
S
witch
© 2006 Network Instruments, LLC
Points of Visibility

For complete coverage, and complete visibility
connect analysis probes to both the core and
access layers at each site
© 2006 Network Instruments, LLC
Phase 2:
Monitoring the Roll Out
Verifying VoIP Health

Cumulative VoIP Metrics
 Satisfaction Scoring
 Aggregate Jitter
 Total Calls
 Codec Verification

Network Configuration and Performance
 Quality of Service / Precedence
 Verifying VLAN Configuration
 Reviewing Link Utilization
© 2006 Network Instruments, LLC
Monitoring Overall VoIP Health
Aggregate Jitter
Call Scoring

High jitter or low
call scoring is an
issue

If this is the case,
go back and
review your setups
© 2006 Network Instruments, LLC
Monitoring Overall VoIP Health
Codecs Used
Total Calls

Is the network
responding as
expected with the
total number of
calls?

Is the right Codec
being used?
© 2006 Network Instruments, LLC
Monitoring Overall VoIP Health

Is Quality of
Service /
Precedence
configured
properly?
QoS
© 2006 Network Instruments, LLC
Verifying VLAN Setup

Identify VLAN
setups and
verify that
VoIP traffic
exists in its
appropriate
VLAN
© 2006 Network Instruments, LLC
Verifying VLAN Setup

Is the
station in its
appropriate
VLAN?
© 2006 Network Instruments, LLC
Link Utilization


Verify utilization
for each link
Ensure that what
you see here
coincides with
information
gathered from
“What-If”
Analysis in the
testing phase
© 2006 Network Instruments, LLC
Phase 3:
Troubleshooting and
Ongoing Maintenance
When Problems Arise

Troubleshooting
 Real-time nature of the call
 Call flow analysis
 Automated problem identification and resolution
 Call mapping for jitter, lost packets, and utilization
spikes
 Use trending data to report on period in question
for traffic analysis

Ongoing Maintenance
 Proactive Monitoring
 Schedule Reporting
© 2006 Network Instruments, LLC
Real-Time Call Analysis


Review calls in
real-time
Track for any
inconsistencies
© 2006 Network Instruments, LLC
Call Flow Analysis



Identify call in question
Track individual stream that comprise the call
Drill down to Connection Dynamics
© 2006 Network Instruments, LLC
Connection Dynamics
© 2006 Network Instruments, LLC
Expert Help

Speed problem
resolution by
obtaining
instant
possibilities of
network issues

Automate
problem
resolution
© 2006 Network Instruments, LLC
Call Mapping


Compare
jitter to
bandwidth
utilization to
understand
RTP/RTCP
response
time
Is this a
bandwidth
issue?
© 2006 Network Instruments, LLC
Trending


Obtain a snapshot of a
time period in question
Check to see if current
conditions are deviating
from historical data
© 2006 Network Instruments, LLC
Monitoring and Alerting

Select which
VoIP
characteristics
should be
continuously
monitored
© 2006 Network Instruments, LLC
Monitoring and Alerting

Determine
what threshold
levels are
acceptable and
set triggers
accordingly
© 2006 Network Instruments, LLC
Monitoring and Alerting

Customize the
appropriate
network action
based on the
event
© 2006 Network Instruments, LLC
Monitoring and Alerting
Shows where
thresholds
were exceeded
Review Expert
thresholds
crossed or
exceeded
© 2006 Network Instruments, LLC
Scheduled Reporting


Customize reports to provide the necessary insight for long-term
analysis and planning
Schedule the reports for automatic delivery on a daily, weekly,
monthly, or even yearly basis
© 2006 Network Instruments, LLC
Sample Report

Choose from
a variety of
report options
and types or
create custom
reports
© 2006 Network Instruments, LLC
Break
Lab 2
Live VoIP
Troubleshooting
Summary
Best Practice #1
Understand and measure
the various components
of call quality
© 2006 Network Instruments, LLC
Best Practice #2
Implement
Quality of Service
Prioritization
© 2006 Network Instruments, LLC
Best Practice #3
Conduct a Site Survey
© 2006 Network Instruments, LLC
Best Practice #4
Deploy analysis tools
strategically for
maximum visibility
© 2006 Network Instruments, LLC
Best Practice #5
Implement VLANs to help
isolate and monitor VoIP
issues
© 2006 Network Instruments, LLC
Best Practice #6
Monitor the rollout to
ensure a positive user
experience
© 2006 Network Instruments, LLC
Best Practice #7
Compare jitter to overall
network bandwidth utilization
to understand response time
© 2006 Network Instruments, LLC
Best Practice #8
Set up your analyzer to
proactively monitor
VoIP activity
© 2006 Network Instruments, LLC
Best Practice #9
Automate problem
resolution
© 2006 Network Instruments, LLC
Best Practice #10
Baseline your network
traffic
© 2006 Network Instruments, LLC
Network Instruments
Solution Set
Analysis Options
Software
Probe
Top Talkers
MultiHop
Analysis
Connection
Dynamics
GigaStor
10/100/1000
Probe Appliance
VoIP Analysis
WAN and Gigabit
Probe Appliances
Gigabit and WAN
Observer Suite System
© 2006 Network Instruments, LLC
Application Analysis
SNMP Management
Customer Feedback
“So far, Observer’s VoIP capabilities
have helped cut down CI Travel’s
phone bill by about 25-30 percent.”
Paul Ingram, CI Travel
© 2006 Network Instruments, LLC
Solid Reviews and Testimonials
“…it is the best packet analysis package we have tested .”
- Dave Bailey, IT Week, December 1, 2005
“Traffic statistics in Observer 11's VoIP Expert tool are more robust, with
call summary, quality scoring and detailed per-call metrics such as
call status, current jitter, call setup, duration, teardown, MOS/R-factor
and QoS levels.”
- Dan Hong, Redmond, November 16, 2005
“Like all Observer features, VoIP Expert is based on the Network
Instruments Distributed Network Analysis architecture, which
means VOIP analysis is available across multiple topologies such
as local-area network, wide-area network, Gigabit Ethernet and
802.11a/b/g.”
- Michelle Speir Hasse, Federal Computer Week, November 21, 2005
Observer continues to receive stellar reviews from
industry pundits and our valued customers
© 2006 Network Instruments, LLC
Recent Wins






Large-scale GigaStor
deployment
Sniffer replacement
For maintaining customer
networks

Large-scale 10/100/1000
appliance deployment across
U.S. locations
Sniffer replacement
For distributed analysis

© 2006 Network Instruments, LLC



Large-scale 10/100/1000 appliance
deployment for 90% of U.S.
locations
Sniffer replacement
For comprehensive visibility
Large scale Expert probe
deployment
For real-time network monitoring
Enterprise Pricing
Enterprise Pricing
Expert Observer
Includes VoIP
US$ 2,895
Observer Suite
Includes VoIP
US$ 3,995
10/100/1000 Probe Appliance
Includes VoIP
US$ 2,495
Gigabit Probe Appliance
Includes VoIP
US$ 11,995
2 TB GigaStor
Includes VoIP
US$ 19,995
4 TB GigaStor
Includes VoIP
US$ 35,000
8 TB GigaStor
Includes VoIP
US$ 50,000



VoIP Analysis included at no additional charge
Shipped with a 64-bit Core, with support for 32-bit systems
Gigabit and WAN Appliances are all 64-bit systems
Network Instruments continues to lead the analysis industry
in performance and value
© 2006 Network Instruments, LLC
Thank You
For more information:
Network Instruments, LLC
Chuck Oxley
phone: 416-285-9191
toll-free: 1-800-526-7919 x3897
e-mail: [email protected]
www.networkinstruments.com