Transcript Future of VoIP: ENUM and number portability

AG Projects
QoS in Next Generation Networks
Telecom Signaling Networks
and Service Forum
January 18, 2006
Amsterdam
AG Projects
I am Adrian Georgescu
QoS in Next Generation Networks
AG Projects
QoS in Next Generation Networks
I work for AG Projects, which is developing solutions for
convergence of the Telecom and Internet
AG Projects
QoS in Next Generation Networks
The opinions expressed in this presentation belong to
myself, my company and most of my friends
AG Projects
QoS in Next Generation Networks
QoS in Next Generation Networks
AG Projects
QoS in Next Generation Networks
Internet is a dumb network, the services (applications)
are performed at the edge
AG Projects
QoS in Next Generation Networks
Dumb means: Internet role is simply to locate and route
packets to the destination IP address
AG Projects
QoS in Next Generation Networks
Internet model is a “best-effort” with no guarantee
AG Projects
QoS in Next Generation Networks
Don’t be fooled by the wording “no guarantee”
AG Projects
QoS in Next Generation Networks
All of you are using it today for your day to day work
AG Projects
QoS in Next Generation Networks
Best Effort is for the Best Reasons!
AG Projects
QoS in Next Generation Networks
Because allowing innovation is what made Internet a
success!
AG Projects
QoS in Next Generation Networks
Ethernet is the most successful network transport
protocol on Earth despite the fact that it does not have
any QoS management built-in!
AG Projects
QoS in Next Generation Networks
Because it could be upgraded to 10, 100, 1000, 10000
Mbit/s without having to change a bit of the routing
protocol or the applications on top of it!
AG Projects
QoS in Next Generation Networks
PSTN is a “guaranteed” transport medium with switched
circuits (paths)
AG Projects
QoS in Next Generation Networks
The question today is:
How do I guarantee a voice call will work with the
expectation that it will not fail or it provides the quality we
use to experience from PSTN?
AG Projects
QoS in Next Generation Networks
- We build IMS, we apply telco logic on top of Internet
- We build and deploy bandwidth management systems,
Pocket Cable, MPLS and others
- Buy our own technology, it has QoS just stick with our
proprietary stuff!
The answers you get today from you vendors:
AG Projects
QoS in Next Generation Networks
Unfortunately, these answers do not address the
problem
They address only your pocket!
AG Projects
QoS in Next Generation Networks
50 millions users of Skype proved that users are
prepared to trade all PSTN “guarantees” with a new
Internet application
AG Projects
QoS in Next Generation Networks
Lets try to find some real solutions which are also cost
effective!
AG Projects
QoS in Next Generation Networks
Internet has two measurable performance parameters:
1. Round-trip time
(RTT) loss (%)
2. Packet
Keep it below 200ms
Keep it under 0.5%
AG Projects
QoS in Next Generation Networks
And two ways to deliver data:
1. UDP protocol (no guarantee of delivery but good for
RT) protocol (guarantee of the delivery, very good for
2. TCP
signaling bad for RT)
AG Projects
QoS in Next Generation Networks
And all these are confined into a bandwidth “force-field”
AG Projects
QoS in Next Generation Networks
Given enough bandwidth, no quality problem exists
AG Projects
QoS in Next Generation Networks
Is cheaper to increase the bandwidth than to build
systems that managed the lack of it
AG Projects
QoS in Next Generation Networks
Do you imagine today a QoS management system for
old dial-up Internet instead of going to broadband?
AG Projects
QoS in Next Generation Networks
Building QoS management systems is difficult, noninteroperable and expensive
By the time such systems are deployed the bandwidth
has doubled or more already, and you are stuck with a
solution build for the past
AG Projects
QoS in Next Generation Networks
Internet performance has been improved dramatically in
the last 10 years
Source: http://www.slac.stanford.edu/xorg/icfa/icfa-net-paper-jan05/
AG Projects
QoS in Next Generation Networks
Complex systems cannot provide a solution for lack of
bandwidth, any system added in the chain will put
strains on QoS as well
AG Projects
QoS in Next Generation Networks
Overall ETSI TISPAN IMS Architecture – all subsystems
Rf/Ro
Ut
Ut
Rf/Ro
Application Servers
Sh
Rf/Ro
ISC
UPSF
Charging
Functions
Dh
Cx
Dx
Iw
SLF
Ib
P3
IMS /
PSTN
Simulation
PSTN
Emulation (R2)
AGCF
Mw
Mw/Mk/Mm
I/S-CSCF
Mk
Mj
Gq'
Mg
MRFC
MGCF
Ie
Network
Attachment
Subsystem
Gm
Gq'
SPDF Resource &
e4
MG
Gq'
Mn
Admission
Control
A-RACF
Re
Mp
Ia
MRFP
SPDF
T-MGF
I-BGF
UE
RCEF
CNG
PSTN/ISDN
P1
SGF
BGF
IP Transport (Access and Core)
Resource &
Admission
Control
Other IP Networks
BGCF
Mr
P-CSCF
Ic
Mk
Mw
e2
IBCF
Mi
P2
IWF
AG Projects
QoS in Next Generation Networks
To improve the quality of service experienced by your
end-users you can follow common-sense best practices:
AG Projects
QoS in Next Generation Networks
1. In the core, monitor the network to avoid congestion, network
loads of 30%-40% are considered adequate for quality of service
and for congestion avoidance
2. At the edge (last mile), provision two paths from the IAD to the
first IP switch.This way RT applications will not collide with non-RT
applications. Invest in last mile bandwidth upgrades instead of QoS
management systems for it
3. In case of packet-loss use smart applications that automatically
fail-over to other hosting center
4. Don’t tunnel signaling and media to home network (like IMS
does), use geographical distribution of resources
5. Use Internet codecs like iLBC or Speex instead of G.7xx
AG Projects
QoS in Next Generation Networks
Remember that:
QoS systems do not create bandwidth!
AG Projects
QoS in Next Generation Networks
This presentation is available at:
http://ag-projects.com
AG Projects
Thank you,
Adrian Georgescu
[email protected]
QoS in Next Generation Networks