International Journal of Electrical & Computer Sciences IJECS-IJENS Vol: 10 No: 03

45

Traffic Engineering with MPLS for QoS Improvement

Prof. Dr. Adeel Akram, Adeel Ahmed Department of Telecommunication Engineering University of Engineering and Technology Taxila Pakistan 47040 adeel.akram@uettaxila.edu.pk, jncip@yahoo.com

Abstract:
o There can be a lot of QoS parameters that can improved in IP MPLS network Backbone such as bandwidth, CPU utilization, memory requirement, resources distribution and database optimization of each router in the backbone. In this paper, we have optimized database of each router by involving DR/BDR election on each interface so that all involved routers use that elected DR router as to communicate hello packets instead of communicating with all routers and hence database size of each router will remain small and require less memory and CPU utilization over MPLS backbone. Simulation results have been verified using MRTG, Olive/Vmware using conventional IP, MPLS without TE and MPLS using TE capability. Keywords: MPLS, VPNs, GGSN, SGSN, IS-IS, OSPF, IBGP, EBGP, RSVP, LDP, BGP, EBGP, DR/BDR. MRTG, PE, CE, P, MPBN o o Yellow arrows is the traffic path from R2 to R3 Indigo arrows shows the traffic path from R1 to R3 Black arrow indicates the under utilization links

1. Introduction
In current IP networks, packets are routed on the bases of destination address and a single metric like hop-count or delay. The drawback of this conventional routing is that this approach causes traffic to converge into the same link; as a result it became a reason for significant increase in congestion and leaving the network in a state of as an unbalanced network resource utilization condition shown in Figure 1.1 [1]

With the implementation of MPLS network and doing traffic engineering we can easily divide the network traffic over the whole network, thus making effective utilization of available links throughout the network. Net result will be cost saving and congestion removal on different nodes in the network by sensing and using the under utilized links [2] In a pure IP network, the shortest path to a destination is chosen even when it becomes more congested. Meanwhile, in an IP network with MPLS Traffic Engineering IGPB routing, constraints such as the RSVP bandwidth of the traversed links can also be considered, such that the shortest path with available bandwidth will be chosen. MPLS Traffic Engineering relies upon the use of TE extensions to OSPF or IS-IS and RSVP. Besides the constraint of RSVP bandwidth, users can also define their own constraints by specifying link attributes and special requirements for tunnels to route (or not to route) over links with certain attributes [3, 4]

2. MPLS Layer Description

Before we discuss MPLS we must know about different protocols features such as OSPF, ISIS, IBGP, LDP and RSVP. In order to describe those features we have taken real networks as an example. Multiprotocol Label Switching (MPLS) is a standards-approved technology for speeding up network traffic flow [5] MPLS involves setting up a specific path for a given sequence of packets, identified by a label put in each packet, thus saving the time needed for a router to look up the address to the next node to forward the packet to. With reference to the standard model for a network (the Open Systems Interconnection, or OSI model),

Figure 1.1 Flow of packets using traditional IP routing [1]

International Journal of Electrical & Computer Sciences IJECS-IJENS Vol: 10 No: 03

46

MPLS allows most packets to be forwarded at the layer 2 (switching) level rather than at the layer 3 (routing) level. In addition to moving traffic faster overall, MPLS makes it easy to manage a network for quality of service (QoS). MPLS have the capability of both Layer-2 and Layer-3 network as shown in the figure 2.1

Figure 3.1 Conventional SGSN/GGSN network diagram OSPF Properties Used: o OSPF is link state protocol means that it will send update to its neighboring routers only when it changes it state. Adjancey States will be Down, Init, Attempt, 2-Way, Exstart, Exchange, Loading, Full Reference BW is always 10^8 OSPF metric is the cost and cost is inversely proportional to interface bandwidth

o o o

Metric = reference BW/ interface BW Router Configuration (OSPF) Example : protocols { ospf { area 0.0.0.0 { interface em3.0; interface em2.0; interface lo0.0; } } } Similar configuration are implemented on all the routers mentioned in Figure 3.1 3.2 MPLS Implementation over SGSN-GGSN network During this implementation we have introduced 8 core MPLS routers in MPLS backbone which are called as P (Provider) routers in the network. Similarly each site contains two M120 routers which are called as Provider Edge routers (PE) in the MPLS Packet Backbone network (MPBN) network. The purpose of two routers is to provide redundancy in each site. In above mentioned Figure 3.1 the four routers in conventional IP network format now act as Customer Edge routers (CE) and these are now uplinked directly with P routers in MPBN BB. We have implemented IS-IS & IBGP between the 8 main core routers of MPLS. OSPF is implemented between PE and CE in MPBN Backbone as shown in Figure 3.2

Figure 2.1 OSI/MPLS Layers comparison

3. IP/MPLS Implementation:
MPLS advantages can be judged with the help of its implementation or replacing conventional IP network with MPLS network. We have implemented MPLS network over existing SGSN-GGSN network which was previously using traditional IP routing scheme. Gateway GPRS serving node (GGSN) 3.1 Conventional IP core on SGSN-GGSN network The conventional IP core was running over serial link (2 Mbps of each maximum capacity) between the 4 routers located at two different places. The links were using OSPF as routing protocol for traffic routing between the two different places as shown in Figure 3.1 [4]

Virtual Routingand forwarding OAM Group Traffic Group vrf_OAM_CG_LI vrf_gngp rd 102:1 rd 101:1

International Journal of Electrical & Computer Sciences IJECS-IJENS Vol: 10 No: 03

47

} interface lo0.0 { level 2 disable; } } } Similarly configuration are implemented on all routers mentioned in Figure 3.2 Router Configuration (IGPB) Example: protocols { Figure 3.2 Migrated network diagram on IP MPLS back bone Please note that MPLS cannot be compared to IP as a separate entity because it works in conjunction with IP and IP's IGP routing protocols. MPLS gives IP networks simple traffic engineering, the ability to transport Layer 3 (Network Layer) VPNs with overlapping address spaces, and support for Layer 2 pseudo wires. Thus routers with programmable CPUs can be implemented [5, 6] explicitly configured hop by hop, dynamically routed by the IS-IS,IGPB algorithm, or Configured as a loose route that avoids a particular IP or that is partly explicit and partly dynamic. [7, 8] bgp { group internal { type internal; local-address 10.0.9.6; neighbor 10.0.3.5; neighbor 10.0.9.7 { } } } Similar configurations are implemented on all routers mentioned in Figure 3.2 Configuring MPLS Traffic Tunnel: [edit protocol rsvp] lab@RT-01# show interface all; interface fxp0.0 { disable; } interface fe-0/0/1.0 { authentication-key "$9$MKZL7Vji.mT3"; # SECRET-DATA } edit@RT-04# show protocols mpls interface so-0/1/0.100; interface so-0/1/1.0; interface fe-0/0/1.0; interface fe-0/0/2.0; interface fe-0/0/3.0; .. Similar configurations are implemented on all routers mentioned in Figure 3.2 4. Results: interface em2.0 { level 2 disable; } interface em3.0 { level 2 disable; By comparing & analyzing the data transfer graphs obtained from the transfer of data across the two networks there is a clear indication that MPLS network with Traffic Engineering has significantly improved the performance of the network.

Router Configuration (IS-IS) Example : // Make all IS-IS speaking interfaces as member of family iso lo0 { unit 0 { family inet { address 10.0.9.6/32; } family iso { address 49.0001.0100.0000.9006.00; } } }

protocols { isis {

International Journal of Electrical & Computer Sciences IJECS-IJENS Vol: 10 No: 03

48

Data rate found much higher in case of IP/MPLS network as indicated by the following MRTG graphs in which we have interface traffic compared with respect to time 24 hours. We have time sampled every millisecond along x-axis and bits per second along y axis. Input and output traffic of each interface is mentioned in MRTG graphs

Delay jitter in conventional IP network can be seen in the following diagram which shows that due to congestion jitter appears

Figure 4.5 jitter rate using conventional IP network Delay jitter in MPLS network without using its TE capability is more or less similar to conventional IP network Figure 4.1 Data rate using IP MPLS TE capability Data rate through conventional IP network is much less than MPLS based network

Figure 4.6 jitter rate using MPLS without using TE capability Figure 4.2 Data rate using conventional IP network Data rate through MPLS network without using TE capability is more or less similar to conventional IP network

5. Conclusion:
In this paper, the simulated concepts behind conventional IP network and MPLS traffic engineering had been verified through various analysis. The outcome of this paper was highly predictable. Through simulation results on (VMWare/Olive) and analysis, it was clearly proven that IP networks only provide best effort service and does not provide any form of resource assurance to traffic flows or any resource guarantees to users. By mixing different types of transport in the same path of the IP network, it can cause network performance problems such as starvation and unfairness for certain traffic. However, with proper MPLS TE applied to the network, the performance of the Network is significantly improved by dividing traffic into separate routes that can be engineered; adequate resources to deliver a good quality of service are achieved. Therefore, the overall network performance is improved in terms of throughput, network utilizations and packet loss rate.

Figure 4.3 Data rate using MPLS without implementation of TE Similarly delay jitter in case of IP/MPLS found almost negligible

6. References:
Figure 4.4 jitter rate using IP MPLS network [1] M. K. Huerta, J. J. Padilla, X. Hesselbach, R. Fabregat, O. Ravelo Buffer Capacity Allocation: A method to QoS support on MPLS networks

International Journal of Electrical & Computer Sciences IJECS-IJENS Vol: 10 No: 03

49

presented at Tenth International Conference on Computer Communications and Networks, 2007 [2] Ibrahiem El Emary, Sherif M. Gad and Ala'a A. Al-Zo'bi New IP QoS Architecture for Voice and Data Convergence over DSL Lines Presented at World Applied Sciences Journal 4 IDOSI Publications, 2008 [3] Victoria Fineberg QoS Support in MPLS Networks presented at MPLS/Frame Relay Alliance White Paper, May 2003. [4] Anton Riedl, Optimized Routing Adaptation in IP Networks Utilizing OSPF and MPLS International Conference on Computer Communications and Networks IETF RFC 2328, April 2008. [5] Ali El Kamel and Habib Youssef An RSVP-TE approach for the End-to-End QoS provisioning within MPLS Domains RFC 4736 International Conference on Computer Communications and Networks, November 2009 [6] Jasmina Barakovic, Himzo Bajric, Amir Multimedia Traffic Analysis of MPLS and nonMPLS Network 48th International Symposium ELMAR-2006, 07-09 June 2006, Zadar, Croatia. [7] Yihan Li, Shivendra Panwar, C.J. (Charlie) LiuPerformance Analysis of MPLS TE Queues for QoS Routing International Conference on Computer Communications and Networks RFC 3031, 2009 [8] Gajeongdong, Yuseong-gu, Daejeon, Multiprotocol label switching & Differential Services as QoS solution IEEE transactions on Networking & Management Vol 12 issue 1, May 2007