MPLS Fundamentals and MPLS L2/L3

VPN Concepts.
BY NAGA R Reddy.

Agenda
1. MPLS definition
2. MPLS is connection oriented or connection less?
3. Why MPLS and Extensions of MPLS
4. MPLS advantages
5. MPLS challenges
6. MPLS header explanation
7. FEC – Forward Equivalence Class
8. Architecture of MPLS and detailed explanation on each individual component
9. Label stacking
10. Label Switching and Label Distribution
11. Label distribution Protocols LDP and RSVP
12. MPLS L2/L3 VPN
13. Signaling Labels
14. Forwarding and Controlling planes

1. MPLS Definition –
a. Multi-Protocol: supports multiple L3 protocols, such as IP, IPv6 and IPX. And use other
layer 2 technologies such as ATM, FR, X.25, Etherenet, etc. MPLS is located between L2
and L3, so they are also called L2.5 protocols.

b. Label: is a short, equal-length, processable information content with partial meaning

only, topology information excluded.

c. Switching: It is not Layer 2 switching. It is English Switching

MPLS packet switching and forwarding are based on labels. For an IP service, when IP
packets enter in the MPLS network, the router in the entrance analyzes the contents
of the IP packet and chooses proper labels for these IP packets. All nodes in the MPLS
 network then depend on these simple labels for forwarding. When the IP packets
leave the MPLS network, these labels are separated by the edge router on the exit.

MPLS is combination of ATM and IP.MPLS is a best practice of considering the good characteristics in
both ATM and IP. Switching characteristics of ATM and flexibility of IP

2. MPLS is connection oriented or connectionless?

For every technology we have control and data plane. For instance, ATM is both control and
connection plane are connection oriented. However, IP technology both CP and UP are connection
less.

When MPLS – Control plane is connection less and User plane is connection oriented.

3. Why MPLS and Extensions of MPLS - MPLS converges the different technologies into
single MPLS network. Various ATM, FR, X.25, Ethernet technologies into converge into 1 single
n/w.

Various Extensions of MPLS:

a) IP/MPLS – IP services , IP forwarding

b) MPLS TP – MPLS transport profile – High bandwidth and protection
c) MPLS TE – Provides QoS
d) GMPLS – Optical Networks

4. MPLS advantages

a) Convergence of several networks into a single network

b) All the TRI-PLAY services (voice, internet and TV) can send over a single transport system.
c) Offering VPNs provisioning, private secure connections between offices irrespective of
geographic locations.

5. Challenges of MPLS

a) QoS – Quality of Service

b) IP duplication– Solution for that IP VPN

6. MPLS Header Explanation

 MPLS Header Length – 4 bytes ( 32 bits)
MPLS label - 20 bits
EXP - 3 bits
Stack - S = 0 /1
TTL - 8 bits

MPLS label – 20 bits short, FIXED length, locally significant and carries the value of MPLS label.
Value varies from 0 to 2^20 - 1. However, 0-15 is fixed. 16 to 2^20 - 1 we can use across the
routers. It is depends the vendor specific. HUAWEI can use 1000

Steps involved in labels -- > Create, assign, distribute

However 15 labels are reserved for special purpose. As of now, only 4 are using.

Any LSR router can’t be assign these labels to any of these reserved labels.

Label 0 – Explicit NULL – Zero Label with QoS

Label 3 – Implicit NULL - no label

Label 1 – Router alert. Closer look on the pocket.

Label 15 – O and M purpose

EXP – COS/QoS
Used to prioritize the pockets, used in queueing and scheduling.

Layer 2.5 Header fields – label, QoS, S, TTL

Layer 2 Header Fields –Ethernet Frame Header - Preamble, Source MAC and Destination MAC, etc.
8021P – COS – 3 Bits

Layer 3 Header Fields – IP V4 header ---Source IP address, Destination IP address, ver, DSCP, TTL, etc
Whereas in IP V4 – TOS (Type of service) 8 bits

Stack
S = 1 , stands for the stack of label is the bottom of the stack
S=0 no more labels on the MPLS.
If multiple labels are attached, it only analyzed the top or outer labels.

TTL
TTL – time to live – it copies from the IP header field when the IP pocket enter into the MPLS rk. It can
specifies the maximum number of hopes that MPLS pocket can travel.

7. FEC – Forward Equivalence Class

The group of same three pockets having the same characteristics (can send the same paths.
If the pockets have same FEC can have the same label.
If the pockets having same label may not be the same FEC. (They might be have different EXP
field values).
Normally, if the pockets are wanted to go same egress edge router would have the same FECs.

8. Architecture of MPLS and detailed explanation on each individual

component
 We can see the different components in the MPLS network. Each router would have different
purpose.

A) CE- Customer Edge router – They do the normal IP routing and send the pocket to the
MPLS Could OR provide edge router
B) LSR – Label Switch Routers - all the routers deployed in MPLS network are called LSR,
which supports MPLS technology. They understand the MPLS labels and transmits the
labels according to that label.
C) Types of LSRs – Ingress LSR (LER) or PE – Receives the unlabeled IP pocket and insert the
label in front of IP pocket and encapsulating into MPLS. Called PUSH operation.
D) Intermediate LSR – Provider router – Receives the labelled packet and performs an
operation on the label usually switching the label on to the next hop. LSR are powerful
deployed in Core network to do high speed switching. This action is called Swap.
E) Egress LSR – LER Label Edge Router OR Provider Edge Router – receives the labelled or
encapsulated pockets and removes the labels called as de-encapsulation. This action
called as POP.

LER routers are very busy and not as powerful as LSR. Moreover, they are not interested in any MPLS
labels because at the end they only send the IP pockets to the customer edge routers.

Therefore, LSR routers ask the second last LSR router to do PHP Operation.
PHP == Penultimate Hop Popping
Penultimate – Next to last

9. Label Stacking

Mpls allows for attaching multiple labels to the pocket. Only first label stack is examined.
When the last label (outer label) is popped then next label becomes active.
Staked labels are used in VPN applications. The first label in the stack is called outer label and
further labels are called inner labels.

For example, here we are considering a VPN for one service and LSP to reach the egress
router. Inner label would be VPN label and outer/top level would be VPN label.

10. Label Switching and Distribution

Basic steps involved in labels:
Create a label
Assign a label
Distribute label

Label Switching
LSP = is unique path created between two LSRs.
LSP – Label-Switched path is unidirectional Path ; enter LSP is in direction. From bidirectional
requires TWO LSPs.

CE sends the pure IP pocket to the ingress LSR. The ingress LSR will do LIFB label forwarding
information base. Each router forms the LSP switching table. That table only informs how to
forward pockets.

The LER router identifies the FEC (Forwarding Equivalence Class) and attaches the label 300
and pushes towards interface 1 towards LSR router.

The MPLS labels are local significance and will change at each point. The LIFB does the ingress
and egress label mapping.
Label Distribution:
We know that Labels are locally significant, that is for only the particular LSR routers.
However, each LSR should know about the labels to learn or use.

Mainly 2 protocols – we can use to distribute the labels

LDP – Label Distribution Protocol
RSVP – Resource reservation Protocol

LDP – can configure easily and don’t provide any control over LSP.
RSVP – complex to configure and provides the control over LSP. It is used where QoS is
needed.

LDP – Works b/w LSRS, no need to have direct connection

LDP doesn’t take any path selection
The path selection done by the IGP protocol (OSPF or ISIS)
 LDP can’t work independently; it should work with combination of any IGP.

The LSR’s use the IGP preferred path as the LSP, so LDP use the labels from that define in this
path.

Finally, how the switching happens with the reference of LIFB table.
We have different tables –
RIB – Routing information base ---- IP table
FIB – Forwarding information base --- IP table
RIB use FIB to forward the traffic
LIB - Label information bas --- MPLS table – Stores the FEC
LFIB – Label forwarding information base
LIB uses the LIFB to forward traffic.

11. RSVP – Resource reservation protocol.

It doesn’t create LSP automatically. LSPs are configured manually. GIVE THE KFC Examples.
RSVP signals LSP’s end to end path

It guarantees the QoS. Sends the RSVP Path messages in downstream direction.
The tail end router returns with RESV protocol along the same path message to ingress router.

Advantages of RSVP
Better utilization, QoS scheme and Link protection.

RSVP can also use the IGP selection. That is the extension of MPLS, MPLS TE.

12.MPLS L2/L3 VPN

VPN Definition - Virtual Private Network, L2 VPN and L3 VPN.

In case L2 VPN – creates a PSW b/w PE and PE

In case L3 PVN only b/w PE and PE routers should make/configure as BGP peers

13.Signaling of labels

Labels are distributed via LDP or RSVP.

If we configure inner labels for PVN that should be read by the edge router. So that signals or
distribution done from LER to LER.

In case L2 VPN - LER can act as a switch and bridges towards CE. The L2 frames are
encapsulated into the MPLS with the inner label from LER to LER used targeted LDP.

In case L3 PVN- LER can acts as a router to the CE. The layer 3 pockets encapsulated into MPLS
 with the inner label to identify the VPN. The inner labels are distributed LERs using an
extended version i- BGP- MP- BGP (Multi-Protocol Broader protocol).

14.Forwarding and control planes

MPLS architecture is divided into two separate components.

Control plane – creating and maintaining the label forwarding information database
Forwarding plane – based on the database it do the forwarding the pockets.

Before pockets can forward, control panel needs to establish all the components necessary
1. Whether MPLS configured globally and other connecting interfaces
2. FECs, LSPs, VPN and LIFB are generated.