Mpls
Mpls
Mpls
Abstract Today’s communication networks and networks. The rollout of MPLS brings the
services are migrating to a converged challenges associated with any new
paradigm centered on IP (Internet networking technology — validating proper
Protocol). MPLS (Multi-Protocol Label conformance with industry standards prior
Switching) has emerged as a key enabling to production deployment and verifying
technology for this migration. MPLS acceptable performance. This white paper
technology has proven its value for provides an overview of MPLS, and Ixia’s
delivering new services while at the same approach to testing and validating that
time allowing migration from old to new technology.
Introduction Several forces shape the current handle a variety of services, both legacy
worldwide communications landscape. and new, over a single network. It enables
One is the general economic slowdown higher-value applications and services to
since early 2000, in particular, the pop of be delivered from the service provider’s
the telecommunications industry bubble. network, thereby reducing requirements
Another is the much-ballyhooed on customer-premises equipment.
convergence of digital communications Integration and consolidation speak loudly
networks (voice, video, data) and the in today’s business environment.
emergence of IP as the protocol of choice.
It’s clear that the migration to MPLS is well
Finally, globalization and deregulation
under way. Every major carrier in the US,
have combined to level the playing field
and many internationally, have deployed or
and increase competitive pressures.
announced plans for MPLS backbones. A
The economic slowdown of recent years 2003 study by Infonetics Research shows
and resulting over-capacity in core 62 percent of service providers are now
networks has forced service providers and engaged in some form of data network
carriers to look seriously at their return on convergence over IP or IP/MPLS, with 86
investment from network assets. With pure percent doing so in 2004. Since legacy
bandwidth becoming, in essence, a services, such as Frame Relay and ATM,
commodity, industry focus has shifted to can be carried over the MPLS network, this
supplying the value-add services network convergence is often transparent
customers need. As new technologies are to the end user enterprise. Moving forward,
adopted, the provider’s ability to newer low cost services such as Ethernet
consolidate disparate existing networks is will drive further adoption.
a key to deploying all services, old and
Beyond large carrier networks, MPLS is
new, profitably. The enterprise market has
also finding its way into the larger
shown a similar response to the slowdown
enterprise networks of organizations such
— increasing efficiencies by pragmatically
as retailers, investment companies,
applying the new technologies that make
government agencies and the military,
such improvements possible.
health care organizations, and technology
Consequently, MPLS has great appeal for enterprises.
telecommunications providers. It can
How Does MPLS MPLS is a technology used for optimizing establish the network topology. MPLS is
Work? traffic forwarding through a network. then overlaid on top of this topology. MPLS
Though MPLS can be applied in many predetermines the path data takes across
different network environments, this a network and encodes that information
discussion will focus primarily on MPLS in into a label that the network’s routers
IP packet networks — by far the most understand. This is the connection-
common application of MPLS today. oriented approach previously discussed.
Since route planning occurs ahead of time
MPLS assigns labels to packets for
and at the edge of the network (where the
transport across a network. The labels are
customer and service provider network
contained in an MPLS header inserted into
meet), MPLS-labeled data requires less
the data packet (Figure 1).
router horsepower to traverse the core of
These short, fixed-length labels carry the the service provider's network.
information that tells each switching node
MPLS routing
(router) how to process and forward the
packets, from source to destination. They MPLS networks establish Label-Switched
have significance only on a local node-to- Paths (LSPs) for data crossing the network.
node connection. As each node forwards An LSP is defined by a sequence of labels
the packet, it swaps the current label for assigned to nodes on the packet’s path
the appropriate label to route the packet to from source to destination. LSPs direct
the next node. This mechanism enables packets in one of two ways: hop-by-hop
very-high-speed switching of the packets routing or explicit routing.
through the core MPLS network.
Hop-by-hop routing. In hop-by-hop routing,
MPLS combines the best of both Layer 3 IP each MPLS router independently selects
routing and Layer 2 switching. In fact, it is the next hop for a given Forwarding
sometimes called a “Layer 2½” protocol. Equivalency Class (FEC). A FEC describes a
While routers require network-level group of packets of the same type; all
intelligence to determine where to send packets assigned to a FEC receive the
traffic, switches only send data to the next same routing treatment. FECs can be
hop, and so are inherently simpler, faster, based on an IP address route or the
and less costly. MPLS relies on traditional service requirements for a packet, such as
IP routing protocols to advertise and low latency.
MPLS Challenges MPLS has made significant progress over continually evolving state and its impact on
the last few years and is well into network performance and scalability.
mainstream deployment in networks
MPLS is not a standalone technology — it is
around the world. But key challenges to
overlaid on Layer 2 technologies such as
attaining more widespread acceptance
Ethernet or ATM, and must operate in
remain. MPLS encompasses a wide range
conjunction with other control plane
of functionality and applications, therefore
protocols, such as IP routing. The
its implementation has an associated high
complexity of MPLS deployments is
level of complexity. Vendors who develop
increased because of this interaction. In
MPLS technology, as well as organizations
some cases, four or more protocols may be
looking at deploying MPLS in a network
involved in a given network scenario,
today, must also factor in MPLS’s
necessitating careful coordination and
Why Test for MPLS MPLS standards and implementations are remain competitive in their market and to
Conformance? dynamic. At the time of this writing, there meet the demands of their customers.
were over 100 IETF drafts associated with Development test and quality assurance
MPLS, and over 20 RFCs. In such a groups therefore need an efficient way to
dynamic environment, standards verify the correctness of their
compliance and the corresponding implementations. Formalized conformance
expectation of equipment interoperability testing against standards supplies this
present significant challenges. confidence. Beyond ensuring product
interoperability and quality, conformance
Equipment vendors find themselves at the
testing can also accelerate product
leading edge of these challenges as they
development by detecting bugs or
continually update their feature sets to the
correcting design issues early in the
latest standards and options, while at the
development cycle, thereby reducing the
same time improving performance and
product’s time to market and hence
scalability. They must do this, both to
increasing profitability.
Why Test for MPLS After verifying the standards compliance RSVP-TE must be verified to determine
Scalability and and interoperability of an MPLS system, the number of protocol sessions that
Performance? the next challenge is to determine the can be sustained.
ability of an implementation to perform in
Together, these numbers will determine
a real network. Given the complexity of the
the quantity of routers that must be
protocols involved and the multi-layered
deployed for a given number of customers.
nature of MPLS, scalability and
performance are often genuine concerns. Performance
Equipment vendors typically test and One of the original proposed benefits of
publish the scalability and performance MPLS was the performance boost
capabilities of their products. End users associated with switching on a label as
will often validate the numbers while opposed to routing on an IP address. While
additionally testing specific network this is of less concern today, the forwarding
scenarios unique to their deployment. performance of PE routers at the edge of
Scalability the MPLS network still involves IP (or
other) lookups and assignments. Vendors
Scalability is typically viewed as the biggest
and service providers alike must test and
challenge in service providers’ MPLS
characterize MPLS devices across multiple
networks today. They must understand the
interface types (Ethernet, POS, ATM) for
dynamics of growth in their networks as
traditional data plane performance
new customers are added, as well as the
metrics:
ultimate limits of their networks. Several
metrics are key in determining scalability: • Data throughput
Test Solution MPLS test tools must be able to perform a extensions to these protocols, for example
Requirements wide variety of functions to test and OSPF-TE and IS-IS-TE, must be supported
validate MPLS devices and systems to allow this aspect of MPLS to be tested.
adequately. For conformance testing, the
Signaling protocol emulation
test solution must be able to fully exercise
the control plane of the device or system MPLS signaling protocols must be
under test. For performance and scalability supported to establish MPLS tunnels and
testing, the test solution must be able to signal L2 and L3 VPNs. Examples of these
emulate MPLS routers at the control plane protocols include LDP, RSVP-TE, and MP-
level and scale up for large capacity BGP. The test tool must be able to run
testing. And it must be able to drive traffic these protocols simultaneously with the
through the system at the data plane level routing protocols on the same network
to fully stress the device being tested. interface.
While simple router emulations can be run Once the MPLS network has been
on PCs or workstations, an optimized test established and all connections signaled,
system must be employed to provide the test tool must be able to inject data
complete testing capabilities and high traffic into the network topology, at speeds
levels of scalability. For example, to up to line rate, and it must be able to
emulate a large network, a network receive traffic as well. This means sending
interface on a test tool must support traffic over all of the LSPs configured on
hundreds or even thousands of IP the test interfaces. On the receiving side,
interfaces and MAC addresses— the test tool must be able to gather
requirements that standard off-the-shelf statistics and capture the traffic for
hardware cannot support. Purpose-built analysis. An increasingly common
test hardware is required to provide the requirement for test tools is to emulate
flexibility and scalability needed to real enterprise applications over the
adequately test MPLS equipment. network. This allows for the
characterization of the network in terms
Routing protocol emulation
the end user really cares about, namely,
The test solution must be able to emulate how their applications will perform.
the full range of routing protocols used in
Automation
today’s networks, including OSPF, IS-IS,
RIP, and BGP. These routing protocols are Since MPLS testing involves complex setup
used to advertise the underlying network and analysis requirements, tests must be
topologies over which the MPLS network is repeatable, which makes automation very
established. In addition, traffic engineering important. Scripting languages are
IX IA IX IA IX I A
IX IA
IX IA IX I A IX IA IX IA IX I A
IX IA
IX IA IX IA IX I A
DUT
IX IA
IX IA IX I A IX IA
Martini
CE encapsulated PE
IXIA L2 traffic traffic I X IA
i
PF c tin
OS asi ar
B M
P ed
LD end
t
Ex
P
PE LD
OSPF
LDP Basic
CE PE
IXIA LDP Extended Martini IXIA
DUT
O
LD LD SP
P P F
Ex Ba
te
n d s ic
ed
M
ar
tin
i
CE PE
IXIA IXIA
MPLS traffic
CE IP traffic PE P PE
OSPF OSPF
IX I A IX I A IX I A
DUT RSVP-TE
MP-iBGP session over
provider backbone
I X IA
I X IA
I X IA
I X IA
emulated
PEs IX I A DUT
P router
I X IA
physical connections
emulated connections
I X IA OSPF sessions
LDP sessions
LDP Extended Martini sessions
Figure 11. IxExplorer LDP Extended Martini session scalability test setup
label information
IX I A
IXIA DUT
detour LSP
ingress LER
IX I A
egress LER 2
Label Information Base (LIB) A table that specifies how to forward a packet
in an MPLS router. This table associates each
label with its corresponding FEC.
Open Shortest Path First (OSPF) A link-state routing protocol used by IP routers
located within a single Autonomous System
(AS) to determine routing paths. MPLS traffic
engineering parameters can be distributed
with OSPF using extensions to the protocol
(OSPF-TE).
Provider Edge Router (PE) A router that operates at the edge of a service
provider’s network, interfacing with the
corresponding Customer Edge (CE) router(s) at
the edge of one or more customer networks.
Routing Information Protocol (RIP) An Internet routing protocol that uses hop
count as a routing metric. RIP is the most
common IGP used in the Internet.