1        Introduction

 

1.1    Thesis Introduction

The pioneer of telecommunication [1], Alexander Graham Bell, invented the telephone in 1876 and in the year 1884 the long distance circuit switched connection was ready for use. Many years later, in 1969, the first military packet switched data network [2] was constructed with few nodes that later increased in size and ended up as the Internet we know today. By the end of 1990, this data network technology became available for the general public. Ever since, it has been a research into new ways for the use of this technology.

 

As it has been described by ITU-T [3], the data traffic is growing at more than ten times the rate of voice traffic. It is estimated that in the near future, data will account for 80% of all traffic carried by telecommunications networks. Therefore, with this rapid change, the past concept of telephone networks, which also carry data, will be replaced by the concept of data networks that also carry voice [3]. Other reasons can be that circuit switched network is less cost effective in terms of network utilization than IP-based network like Internet and new services, which need both voice and data transmission simultaneously. An evolution from simple document sharing and sending e-mail to using the Internet for real-time voice, video and entertainment is causing a convergence of the circuit switched telecommunication network and the IP-based network. Thus, the telecom industry has begun their task for using IP as the bearer of traffic.

 

This has of course consequences for the best-effort services that Internet was intended for. Internet and other IP-based networks have not been able to guarantee low latency and reliable packet delivery at the low delay that is needed for services like real-time voice communication. There are possibilities to implement for example Integrated Services to accomplish this quality, but this gives no reliability when failure in the network occurs.

 

The Internet has grown geographically, and the increase in number of hosts and traffic volume in its turn increases the operational efforts. With this in mind and the framework for the convergence explained above, a need for a more cost effective and reliable network technology has emerged. Cisco developed “tag switching”, which became a forerunner for the new technology, placed between layer 2 and 3 in the IP stack, and it was finally called Multiprotocol Label Switching (MPLS) [4]. MPLS has been a part of IETF since 1997 [4] and for ITU-T since second quarter of 2000, and these organizations will play key roles in the further developing of the MPLS technology [3]. A study of tier-one and tier-two in carriers in U.S by Infonetics Research shows that respectively 56% and 65% of them planned to implement MPLS in 2001 [4].

 

IETF, ITU-T and others develop OAM mechanisms for MPLS to secure failure detection and operation of the network. These OAM mechanisms are, in this thesis, to be compared to those in IP. Label Switch Path (LSP) connectivity [21] is used to verify that LSPs maintain connectivity and tells affected routers about failures. Another example of LSP connectivity functionalities is MPLS Ping [5]. The OAM packets traverses along the LSPs and a balance between OAM traffic and work traffic must be maintained. IP uses ICMP [18a] to advertise failures, but the MPLS architecture does not provide a similar mechanism.

 

There exist proposals for different types of fast rerouting (read more at [35] [7]) and protection switching (read more at [32] [14]) for MPLS. These properties, which do not exist in IP, give the ability to switch quickly over to another LSP when failure on the working LSP has occurred.

 

Traffic engineering (TE) on MPLS [31] gives network operators significant flexibility in controlling paths of traffic flows, that traverse across their networks. TE allows policies to be implemented that can optimize the performance of networks. Such possibilities are currently not available on IP.

1.2    Thesis description

This thesis shall evaluate OAM for MPLS networks. The principles with dedicated OAM cells will be compared to the use of other OAM mechanisms existing at the IP layer (e.g. SNMP), and MPLS OAM principles (IETF and ITU-T) shall be evaluated. The thesis shall propose mechanisms that can be recommended for large backbone networks.

 

A study of ongoing activities within ITU-T and IETF is required. A testbed should, if feasible, be set up to perform necessary tests. A testbed with software routers (PCs) is recommended if this is implementable. If a testbed is to be used, a study of available software for simulating such routers must be performed. It will probably be necessary to write some additional software to insert the necessary OAM cells, as this probably doesn’t exist from any vendor yet. However, if the time shows that the use of a testbed is not implementable, the thesis will be performed theoretically.

1.3    Thesis progress

During most of the time that we have been studying OAM on MPLS, we have been thinking of new properties for OAM on MPLS. We have come up with some new aspects on this field concerning OAM mechanisms.

 

The main point of this thesis is to compare MPLS OAM cells to the existing at the IP layer, thus this thesis is not discussing what link layer protocols to use in a large backbone networks. Therefore, the link layer protocols ATM and Ethernet are not discussed in detail.

 

Much work was laid down in finding suitable testbed architecture and applications for allocation of packet stream, and packet sniffing. Correspondence through mailing lists was also made to exchange ideas around the testbed. The testbed was to be carried out in the beginning of April, but due to complications at Cisco we had to postpone this testing. In the middle of May the complications were still not solved, and therefore we had to omit the testbed.

1.4    Literature review

Lately the telecommunication industry has been highly focused on how their leap towards using IP for telecommunication services. We expect that MPLS may be chosen for the bearer of IP in future large backbone networks, and that the OAM mechanisms of these backbones will be important. The current work at the Internet Engineering Task Force (IETF) in the draft Fast Rerouting [35] reveals an active working environment for OAM mechanisms on the MPLS platform. This fast rerouting protocol is originally intended for link layer errors, whilst the Protection Switching [32] at the International Telecommunication Union Telecommunication Standardization Sector (ITU-T) shows that rerouting can also be applied to the full LSP using various solutions. This is also a part of the OAM requirements that may be proposed for OAM functionality in MPLS networks [20] by the ITU-T. ITU-T has also been discussing various OAM mechanisms for MPLS [21] lately.

 

The work on OAM standardisation is still in progress, and is highly prioritized in scientific e-mail based discussion groups, both at ITU-T and at the MPLS Resource Center [49].

 

We have also used several books in this thesis. One of them is Computer Networks: A System Approach [18], written by Larry L. Peterson and Bruce S. Davie. This book has increased our learning on IP in general. On the MPLS area, the book MPLS Technology and Applications [13], written by Bruce S. Davie and Yakow Tekhter, have provided us with a general introduction to MPLS.

1.5    Report outline

This thesis should not be seen as a work reference or encyclopedia, but rather be seen in its entirety, where most pieces of information can be traced back to the starting chapter; giving valuable information as a whole.

 

Chapter 2 gives the reader understanding in terms of the OAM, backbone, MPLS and IP technologies, and providing a basis for latter discussion.

 

Chapter 3 describes OAM functions of these technologies and classifies them according to their functions, before we go into details of comparing the OAM mechanisms on MPLS to existing OAM mechanisms on IP in Chapter 4.

 

In Chapter 5 we provide our recommended mechanisms and new ideas at the OAM level. Fortunately, one of these ideas may be patented, thus we had to move most of this information to Appendix D. The content of this idea is restricted, and may later be published to the general public. This information is of course still available to the censors.

 

When it comes to references, it may be helpful for our readers to clarify how we have been using them in our text. Firstly, we have referred wherever possible. Even when we have altered the text, or provided sniplets from various sources, we still have given credit in form of a reference to the owners of the idea or the information. The references are a numeric number inside brackets like [“number”]. Sometimes, we have written a specific line that contains a number at the end before a period. This means that the above text is referred. When the content of a paragraph is referred from a single source, we give reference by providing a number in brackets after the period.

 

Now we have given some information about the information and ideas provided by this thesis, now it’s your turn to take a dive into the world of OAM.


MECHANISMS FOR OAM ON MPLS IN LARGE IP BACKBONE NETWORKS (c) 2002 Hallstein Lohne, Johannes Vea, a graduate thesis written for AUC/ERICSSON