white paper

284 23-3123 Uen Rev A | January 2009

Seven reasons to use

end-to-end thinking when
building all-IP networks

Most mobile operators around the world are already well on their
way to all-IP networks. Their concern now is how to complete that
transition and handle the substantial growth now being seen in
mobile broadband services. They can address these concerns by
choosing an end-to-end IP transport solution.
 Introduction
The benefits of an all-IP network are clear. operators already have some IP infrastructure
Ethernet-based IP transport, for example, deployed to deliver GPRS traffic but this has
reduces costs and enables service to all not been updated sufficiently. This
users, from low-revenue voice-plan-only infrastructure often uses older technology
users to intensive users of high-speed data and prevents operators from gaining the full
services. Security and QoS (quality of service) benefit of their IP investments.
can be built in as part of the network design, Operators are advised to take an end-to-
providing resiliency and high availability on a end approach. This means taking a holistic
par with traditional circuit-switched telephony view of both service requirements and how
networks. An all-IP network meets future these requirements affect the structure,
requirements, solving data and capacity technology and network components of
bottlenecks and preparing the way for the the optimal solution for each part of an
Evolved Packet System (LTE and SAE). operator’s network. When possible,
The question many operators face is how operators should choose a single vendor
to make an efficient transition from their that understands the entire network from a
current voice-centric transport infrastructures telecom perspective and can deliver
to a fully packet-data-centric model. Each telecom-specific goals such as high
operator is starting from a different position. capacity, security, QoS, manageability,
Many mobile operators’ networks have been resilience and a future-proof migration
built in an ad-hoc manner or grown through path. Building an all-IP network is more
acquisition, resulting in high opex, lower than simply connecting a number of
reliability and increased round-trip delay. In routers, switches and firewalls.
the core network, for example, many

2 Seven reasons to use end-to-end thinking when building all-IP networks Introduction
Mobile network overview

Figure 1: Mobile network functional segmentation

Before analyzing the benefits of an end-to- GGSNs) controlling data sessions and traffic
end solution, it is worth defining the parts of forwarding as well as MSC and MGw
the mobile network and how they interact functionality for providing call switching and
with each other. associated services.
The entire infrastructure of a typical
mobile operator can be divided into several Mobile backhaul: Within the RAN domain,
distinct parts. the mobile backhaul takes care of
transporting traffic between the RBS and
RAN Domain: This is the access/aggregation BSC and/or RNC sites. There are many
part, funneling traffic from a large number of ways in which an operator can do this,
RBSs towards a much smaller number of and the RAN solution should be
RNCs (for WCDMA traffic) or BSCs (for GSM independent of the transport method
traffic). Voice traffic is directed towards the chosen. Possible transport networks
core network via Media Gateways (MGw) include L2 networks (Carrier Ethernet), L3
whereas packet-data traffic is directed networks (BGP/MPLS L3 VPN) and IP over
towards SGSNs and GGSNs. E1/T1 using MLPPP.

Mobile core domain: The core network sits In cases where the chosen transport network
between the radio access networks and is not immediately available at the RBS site,
external networks such as the internet, last-mile access must also be considered.
PSTNs and other mobile operators. It Possible solutions for this include microwave,
contains the serving nodes (SGSNs and copper and optical fiber (FTTx).

Seven reasons to use end-to-end thinking when building all-IP networks Mobile network overview 3
 Figure 2: Mobile backhaul for the radio access network

Mobile backhaul can be divided into the operator could have a huge number
“low” and “high” RANs (LRAN, HRAN) of RBS sites concentrated towards
reflecting the asymmetrical nature of a much smaller number of switch
backhaul networks, in which a large (RNC/BSC) sites.

4 Seven reasons to use end-to-end thinking when building all-IP networks Mobile network overview
Reason 1: reduced opex, capex
and TCO
An all-IP network will mean significant savings working together efficiently, from the
for an operator’s bottom line. IP-based radio access through the backhaul to
transport is not only cheaper on a bit-per- the core network. This should be done
second basis but is more easily scaled, which with the minimum of overlap or
allows operators to better accommodate underutilized infrastructure.
the growing demand for higher bandwidth. An end-to-end approach will:
The simplicity of the Ethernet solution also ✒ Use common platforms and technologies
reduces management and deployment for both the radio-access and core-
costs substantially. network elements to minimize both
There is obvious potential for considerable capex and opex as well as enable
cost reduction but how can an operator faster integration
best exploit it? ✒ Reduce capex by replacing existing
With a holistic approach, operators and multiple-transport networks with a single
system integrators evaluate every network integrated IP network
element as well as the management systems ✒ Achieve lower total cost of ownership
that make up the entire communication (TCO) for the backhaul through efficient
chain. The optimal solution should be aggregation regardless of the physical
designed to scale and adapt easily. The infrastructure used (optical, copper or
operator will need to get all the elements microwave media).

Reason 2: shorter time to market

and more efficient use of current
infrastructure
Getting products and services to the market to maximize their investment in the
first is the aim of any efficient business existing systems.
organization. However many operators have No two operators are alike but one thing is
inflexible networks and management systems common to all: the need both to maintain their
that hinder rather than support the launch of ongoing business throughout a network
new services. Missing a market window transformation, and to reduce costs by
due to rollout delay or failure can have utilizing the current infrastructure as much as
serious consequences for an operator’s possible. An end-to-end solution should be
revenue stream. implemented in step with an operator’s
Operators need a network solution that is changing needs and business plan;
verified and proven – one that has investments must be made when they
undergone lab testing as well as integration make sense.
testing with the core network, radio-access- An end-to-end approach will:
network equipment and other appropriate ✒ Utilize network and service-management
services. This approach should also look systems that can simplify the launch of
at the existing network and allow operators new services by simplifying customer

Seven reasons to use end-to-end thinking when building all-IP networks Reason 1, 2 5
 handling, service provision and RBS and BSC/RNC sites. In this way, an
resource management operator planning to move to IP can start
✒ Allow the backhaul network to evolve to IP the migration process by deploying an all-IP
RAN when the capacity requires it network on its legacy infrastructure
✒ Allow operators to use their existing ✒ Be standards-based for successful
transport networks, previously used for interoperability with existing network
TDM, to transfer IP packets between the equipment and transport services.

Reason 3: increase network

security
Security is a growing issue for carriers and transport network (sometimes using public or
operators. As mobile networks continue their semi-public networks) are the two main causes
evolution toward all-IP, it is crucial to consider for concern. It is much more difficult to add
network-wide security at the planning stage, security after the network is operational, and
rather than as an afterthought. Increased this may require a major network redesign.
connectivity thanks to the possible use of An end-to-end solution will:
public networks for backhaul and increased ✒ Build on a strong set of basic security
adoption of native IP devices both make features and inherently secure products
networks more vulnerable. that help operators lower their capex by
A comprehensive security policy – having fewer dedicated security nodes
developed according to the security needs of and lower their opex by reducing the need
the operator – should be a central pillar of any for administrative processes
end-to-end approach. In the network, a risk ✒ Allow basic or advanced levels of
analysis will be made for every node and traffic protection according to the types of
type. Separate security zones will be created services provided and the operator’s
within and between sites and then protected security policies
according to their assessed risk levels. ✒ Ensure that every element in the
In the radio access network, increased RBS network can survive a denial of service
connectivity as well as insecurity in the (DoS) attack.

6 Seven reasons to use end-to-end thinking when building all-IP networks Reason 3
Reason 4: guaranteed quality of
service (QoS)

Figure 3: Guaranteeing end-to-end QoS across multiple network domains

QoS is the ability to guarantee a certain level An end-to-end approach ensures telecom-
of performance for traffic through the network, grade QoS by using:
including throughput, delay, delay variance ✒ Network dimensioning – including a
and reliability. detailed inter-site traffic matrix for the
Different traffic types require different levels different traffic-class types. The network-
of QoS. For instance, network control and dimensioning process results in
management traffic will place high demands bandwidth in the network being
since the network depends on them, whereas provisioned according to transport needs.
best effort data traffic such as web browsing ✒ Traffic differentiation – determining as a first
could have substantially reduced priority. step all traffic types traversing the network.
Without a well defined QoS policy, an Those traffic types can then be grouped
operator’s network will function much less based on the delay sensitivity of the traffic.
efficiently, creating delays in time-sensitive ✒ Queuing and scheduling – determining
services and resulting in dissatisfied the order of transmitting packets of the
consumers. QoS can conversely be a revenue forwarding classes on the outgoing
generator, as consumers become willing to interfaces. Each router and switch will
pay for better QoS such as tiered services or implement internal queuing and
guaranteed throughput levels. scheduling algorithms to ensure the QoS
An end-to-end approach will ensure QoS of the different traffic.
requirements are already fulfilled at the ✒ Admission control – required to ensure
planning stage. The solution consists of QoS that infrastructure capacity is not
mechanisms in mobile terminals, radio access overloaded. Mobile nodes, such as the
networks, core network, service network and MGw, will limit the traffic injected into the
IP backbone. QoS functionality in gateway backbone network, thus maintaining
nodes enables interworking with other external voice quality when there is congestion in
networks. The solution will encompass all the backbone network. The network will
network layers from top to bottom, as therefore behave in the same way as
well as every network element from an traditional TDM-based networks when
end-to-end perspective. congestion occurs.

Seven reasons to use end-to-end thinking when building all-IP networks Reason 4 7
 Reason 5: high availability
Traditionally, IP-based networks had a To succeed, it is vital that an end-to-end
reputation for failing to meet the high reliability approach is taken right from the beginning. It
standards associated with TDM and ATM is also important to understand what the
networks. However improvements in exact network requirements are – for
IP-network-node software and hardware has instance, how quickly a failover between two
made goals such as 99.999 percent uptime paths needs to be carried out to guarantee
possible, as well as quick failover to redundant that users will not be affected.
network paths in case of any single failure in An end-to-end approach will ensure:
the network. ✒ High availability of individual nodes –
The key to building resilient networks is to achieved through hardware redundancy
understand that reliability must be supported and an operating system designed for
by all the network elements, from the services high reliability
and applications right down to the platforms ✒ Redundancy of nodes and links – a
upon which the network operates. network designed so that no individual
All nodes (mobile nodes as well as link or node failure can prevent traffic
IP-network nodes) included in the packet from reaching its destination
transport will have built-in resiliency ✒ Mechanisms for rapid detection of path
mechanisms, some using standard failure, and mechanisms for rapidly
routing protocols, some using built-in moving traffic onto alternative paths such
application-based resilience mechanisms as routing protocols, host resilience
and others using host-based resiliency mechanism and protection protocols in
mechanisms. It is not always easy to the backbone network
understand which mechanisms to choose ✒ Security – because many network failures
to achieve the best possible result in a are caused by a lack of sufficient
large network with many traffic types and security, leaving them vulnerable to DoS
traffic nodes. attacks and similar threats.

Figure 4: Reliability and security place demands at every level

8 Seven reasons to use end-to-end thinking when building all-IP networks Reason 5
Reason 6: assured migration path
When it comes to any technology, knowing intermediate steps towards the 3GPP
what is around the corner is never easy. In the standards for SAE and the Evolved Packet
telecom world the situation is made Core (EPC) – the specification for changes to
somewhat clearer through agreed standards the packet-core network architecture. These
that map out the progression and evolution of can improve network scalability, make future
the mobile network. In the coming few years, expansions more cost-effective and reduce
both GSM/UMTS and many CDMA operators transmission costs.
will move to the new mobile standards known An end-to-end approach will:
as Long Term Evolution (LTE) and System ✒ Minimize future capex and opex by
Architecture Evolution (SAE). LTE will have defining clearly where investments
significant benefits over today’s networks should best be made to meet the future
including more efficient use of the radio needs of LTE/SAE
network, faster transfer rates, higher mobility ✒ Make sure infrastructure is capable of
and lower latency. supporting the connectivity and capacity
By adopting an end-to-end methodology requirements of the future (high-density
when moving to an all-IP network, operators gigabit Ethernet, scalability, forwarding
can better prepare themselves for the future performance, advanced services – deep
steps that will be needed when moving to packet inspection, policy, subscriber
LTE/SAE. Operators can now take management and so on).

Reason 7: more powerful and

efficient network management
It is clear that any network transformation service-level assurances are met for both
should be matched by a corresponding contractual purposes and end-user
network-management transformation. service satisfaction
Management transformation can help ✒ Automate some network-deployment
decrease costs, ensure high-quality service functions with auto-configuration,
delivery to customers, maximize revenues common configuration templates and the
and establish a sound foundation when simplification of complex tasks such as
moving to next-generation networks MPLS LSP configuration
and services. ✒ Simplify management systems by
An end-to-end approach will help create incorporating capabilities within
effective management systems and each network domain by utilizing
processes across the entire network; this will a network-domain manager to make
enable operators to attract and retain the capabilities and features of each
customers. It will also increase operational network domain available to all other
readiness, allowing a shorter time to market management domains
for new products and technologies. ✒ Reduce dependency on stand-alone
An end-to-end approach will: vertical systems
✒ Monitor key performance indicators ✒ Reduce operational expenses and move
(KPIs) to guarantee the health of the to cost-effective, integrated management
network and track key criteria such as based on an architecture with clear roles,
network loading. This ensures that responsibilities and interfaces between
capacity planning is effective and that management domains.

Seven reasons to use end-to-end thinking when building all-IP networks Reason 6, 7 9
 Conclusion
Transition is never easy. Operators today face provide guidance for the entire network, and
transition not only in their networks and its professional services staff can help
management systems but also in their analyze, identify and recommend when to
organization and processes. Transition, increase backhaul capacity, when to
however, can be made easier if an operator aggregate traffic, and which ATM or TDM
knows where it wants to go and can take a components should be replaced with IP-over-
comprehensive view of its entire business Ethernet components and when.
and network landscape. Telecom quality is the fundamental
This requires an IP-infrastructure solution differentiator. It is not only about reliability
that is coherent, reliable and cost-effective. A but also standards compliance and a
new network must be able to accommodate commitment to support solutions as
enormous growth in broadband packet entities rather than a collection of boxes.
data, largely driven by video or multimedia, Solutions must have committed roadmaps,
and be capable of continuing the evolution development teams and dedicated
to all-IP and the Evolved Packet System support. This is why it is essential to
(LTE/SAE) without extensive reevaluation partner with a major telecom industry
and investment. player that understands and is helping
Choosing a single vendor with expertise in shape future mobile industry standards,
all the elements of the mobile network can and can guide operators through the
bring enormous benefits. A single vendor can challenges ahead.

10 Seven reasons to use end-to-end thinking when building all-IP networks Conclusion
Glossary
2G second-generation radio technology PLMN public land mobile network
for wireless networks PSTN public switched telephone network
3GPP 3rd Generation Partnership Project QoS quality of service
3G third-generation radio technology for RAN radio access network
wireless networks
RBS radio base station
AGW application gateway
RNC radio network controller
ATM Asynchronous Transfer Mode
SAE System Architecture Evolution
BGP Border Gateway Protocol
SGSN Serving GPRS Support Node
BSC base station controller
SLA service level agreement
BSS business support systems
TCO total cost of ownership
capex capital expenditure
TDM time division multiplexing
CDMA code division multiple access
UMTS Universal Mobile
DRM digital rights management Telecommunications System
DoS denial of service VPN virtual private network
DPI deep packet inspection WCDMA Wideband Code Division
EPC Evolved Packet Core Multiple Access
FTTx Fiber to the x, where x can be WLAN wireless local area network
N (node), C (curb), B (building)
or H (home)
GGSN Gateway GPRS Support Node
GPRS General Packet Radio Service
GSM Global System for
Mobile communications
HRAN high radio access network
HSPA High Speed Packet Access – part of
3GPP WCDMA standard
IP Internet Protocol
ISDN Integrated Services Digital Network
ISP internet service provider
KPI key performance indicators
LRAN low radio access network
LSP Layered Service Provider
LTE Long Term Evolution
MGw Media Gateway
MLPPP Multi link Point-to-Point Protocol
MPLS Multiprotocol Label Switching
MSC mobile switching center
opex operational expenditure
OSS operations support systems

Seven reasons to use end-to-end thinking when building all-IP networks Glossary 11
 References
✒ Ericsson AB, Ericsson Review 2004. Modularity is key when designing packet backbone
networks for mobile services. Available at: http://www.ericsson.com/ericsson/corpinfo/
publications/review/2004_01/182.shtml [Accessed 12 January 2009].

✒ Ericsson AB, Ericsson Review 2004. Security architectures for mobile networks.
Available at: http://www.ericsson.com/ericsson/corpinfo/publications/
review/2004_02/191.shtml [Accessed 12 January 2009].

✒ Ericsson AB, Ericsson Review 3 2008. Mobile broadband backhaul: Addressing the
challenge. Available at: http://www.ericsson.com/ericsson/corpinfo/publications/
review/2008_03/ [Accessed 12 January 2009].

✒ Ericsson AB, white paper. March 2006. Telecom quality in all-IP networks. Available at:
http://www.ericsson.com/technology/whitepapers/3063_Telecom_Quality_all_IP_B.pdf
[Accessed 12 January 2009].

✒ Ericsson AB, white paper. October 2008. High speed technologies for mobile backhaul.
Available at: http://www.ericsson.com/technology/whitepapers/broadband/high-
speed_mobile_backhaul.shtml [Accessed 12 January 2009].

12 Seven reasons to use end-to-end thinking when building all-IP networks References