INTERNET-DRAFT Subrata Goswami Expires February 12, 2003 Independent Consultant Sept 13, 2002 A Simple Analysis of Mobile IP v4 Status of this Memo This document is an Internet-Draft and is in full conformance with all provisions of Section 10 of RFC2026. Internet-Drafts are working documents of the Internet Engineering Task Force (IETF), its areas, and its working groups. Note that other groups may also distribute working documents as Internet-Drafts. Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet- Drafts as reference material or to cite them other than as "work in progress." The list of current Internet-Drafts can be accessed at http://www.ietf.org/ietf/1id-abstracts.txt The list of Internet-Draft Shadow Directories can be accessed at http://www.ietf.org/shadow.html. The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in RFC 2119 [RFC 2119]. Abstract This document analyzes Mobile IP v4 from a deployment point of view. With the popularity of wireless LAN technologies subnet roaming has become a prominent issue. Some critical scalabilty issues for Home Agent and Foreign Agents are pointed out. 1. Overview and Rationale In the 802.11 wireless lan [WiFi] network an 802.11 client connects to an 802.11 Access Point (AP) at the link level.Mobile-IPv4 (MIP4)[MIPv4] enables a Mobile Node (MN), equiped with 802.11 access card, to roam from subnet to subnet. At present MIP4 is the most mature and widely implemented standard for network level mobility. Hence it makes sense to see what are the issues that may limit its usefulness from a practical perspective. 2. Number of Tunnels A hypothetical Mobile IP network is shown in the following figure. 9 subnets are shown in the figure. Each subnet has one Home Agent (HA) and one Foreign Agent (FA). A Mobile Nodes (MN) would have their home network in one of these subnetworks. The FA and HA could be seperate entities or may be the same entity (i.e. have the same IP address). ------------------------------------- | [FA0]| [FA1]| [FA2]| | [HA0]| [HA1]| [HA2]| | | | | |[MN11] | | | |[MN21] | | | ------------------------------------- | [FA3]| [FA4]| [FA5]| | [HA3]| [HA4]| [HA5]| | | | | | | | | | | | | ------------------------------------- | [FA6]| [FA7]| [FA8]| | [HA6]| [HA7]| [HA8]| | | | | | | | | | | | | ------------------------------------- [MNij] - Mobile Node j in Home Network i [FAi] - Foreign Agent in network i [HAi] - Home Agent in network i Figure 1: Mobile IPv4 deployment over an address space of 9 sub-nets. A tunnel needs to be created when a single MN roams into a foreign subnet. If there are Sn subnets, then there can be Sn*(Sn-1) or Sn*(Sn-1)*0.5 tunnels depending on when FA and HA are distinct or the same entities respectively. Thus tunnel configuration and management is an n^2 problem. For the example in Figure 1, there would 9*8=72 or 36 tunnels. A large organization that has an Class-A (or /8) address space can have 2^16 subnets of 254 nodes. That can potentially imply 2^16*(2^16-1)*0.5 = about 2 billion tunnels. Although, at any point of time only 2^16*254 = about 16 million of the tunnels would be active. For an organization with Class-B (or /16) address space, those numbers would be 2^8*(2^8-1) and 2^8*254 or about 64,000 and 64,000 respectively. 3. Home Agent Throughput The second scalability issue is the throughput of the HA. If the larget subnet has Nn nodes, then the HA of that subnet may potentially have to support Nn tunnels. Now if each node has bandwidht requirement of b bits per second, then the HA needs to have a throughput of bNn. Thus there is a linear relationship between an HA and the number of nodes in the subnet. For example, in a 1022 node subnet with 1.0 Mbps per node demabd, the HA needs to have a throughput of 1Gbps. 4. Foreign Agent Throughput The third scalability issue is the throughput of the FA. If the average number of nodes is An nodes per subnet, then there is the pathological case of all hosts of the Sn subnets roaming into one subnet. In such a situation the FA would have to have a throughput of bAnSn. To get an idea of what this means, let us consider a Class B space with 254 node subnets. If each MN demands 1Mpbs, then the FA would need to pass 254*255*1= about 64 Gbps. Although, in most real situations only a fraction of the nodes would roam into one subnet. In the pathologocal case mentioned above, the number of tunnels that needs to be terminated by the FA is approximately the number of nodes available in the adderss space. If the pathological case needs to be supported,then it might be worhwhile to have multiple FA's per subnet so that the load can be sahred. This in the extreme case would result in co-located FA , which has severe implications in the number of IP addresses used. 5. Mobile Node Issues The primary impact on an MN in the pathological case mentioned above would be inability to register (hence no service ) with an FA, if the Visitor Table in FA is limited. Even if the Visitor Table supports 64,000 entries, the FA would still need to add, delete, and search a large table. 6. Acknowledgments All the RFC's, IDĘs, freely available 802.11 standards, and Linux web-sites. 7. References [MIPv4] Perkins, C., "IP Mobility Support", RFC 2002, October 1996. 8. Author's Address Subrata Goswami, Ph.D. Independent Consultant Newark, CA 94560 sgoswami@umich.edu This document expires February 12, 2003.