270 likes | 441 Views
VLSM and RIPv2. Warren Toomey GCIT. Introduction. Switches deliver frames to stations. Routers deliver packets to networks. Only the final router has to deliver a packet to a station, as they are both on the same network.
E N D
VLSM and RIPv2 Warren Toomey GCIT
Introduction • Switches deliver frames to stations. • Routers deliver packets to networks. Only the final router has to deliver a packet to a station, as they are both on the same network. • Routers operate at OSI Layer 3. In particular, IP routers use link-independent addresses. • An IP address identifies an interface which is connected to a network. This means that routers have at least 2 IP addresses. • This also means that every point-to-point WAN connection between routers is also a separate IP network.
Introduction • Why do routers only route to networks? To keep their routing tables small, and to keep the process of routing fast. • At present, hundreds of millions of Internet stations connected to the Internet. But only hundreds of thousands of networks connected to the Internet. • Router tables look like the following:
Networks and Subnets • Originally, 3 network sizes: Class A, B and C • Problems: not given out in a geographical basis • No ability to aggregate networks • Sizes are generally too big or too small • Each network is a broadcast domain: broadcast storms
Subnets • Earliest solution: subnet a network into many equal-sized networks • Subnets kept hidden behind router, which aggregates all subnets into one network
Subnets • Done by borrowing bits from the host field to create the subnet field. Smaller subnets = smaller broadcast domains • Problem: R1 to R2 link is a subnet: we only need 2 IP addresses but there are 254, so we waste 252 IP addresses
Routing with Subnets • Early routing protocols (e.g. RIPv1) were classful but could deal with subnets as long as they were contiguous (not separated). • In previous diagram, R1 could advertise 131.66.0.0 /16 • Discontiguous subnets not possible. R3 gets same advertisement from R1 and R2. Which way is right? Both of them!
Drawbacks of Subnets • All subnets are the same size. Many IP addresses are wasted • Cannot deal with discontiguous subnets, as any routing advertisement has to be classful (only /8, /16 or /24)
CIDR: Classless Internet Domain Routing • Create new routing protocols, or update existing routing protocols, to send the subnet mask with every advertisement • Allows the receiving router to determine the size of the network, i.e. the range of host addresses on that network • Allows routers to aggregate routes and advertised summarized routes to their neighbours • e.g. two adjacent /24 subnets can be advertised as one /23 subnet
CIDR: Classless Internet Domain Routing • RIP was updated to RIPv2 to support CIDR • New protocols such as EIGRP and OSPF support CIDR • IP address allocations no longer need to be contiguous or the same size, as they were with subnets
Netmasks, Network Sizes, Step Factors • Netmask used to indicate the network part of the IP address (used to route packets) and the host part (used by last router for final delivery to station) • Traditionally, a 32-bit pattern with 1's for the network part and 0's for the host part, e.g. • 11111111 11111111 11111111 00000000 • Dotted decimal notation: 255.255.255.0 • Prefix notation: /24 (number of 1 bits) • We will use all three netmask formats
Netmasks, Network Sizes, Step Factors • Netmask tells you the size of the network • e.g. /24 has 8 bits for hosts. 28 = 256 hosts • On each subnet, 2 IP addresses are reserved • Netmask also tells you the “step factor”: how far apart the subnets are apart • Once you have the base IP address of a network plus the mask, you can determine: • The network identity (same as base address) • The broadcast address (last IP address in range) • The set of usable IP addresses (all but 1st and last)
Netmasks, Network Sizes, Step Factors • The number of addresses on any network is always 2host bits, and the number of usable host addresses on any network is always 2host bits -2. • The network identity address (i.e the first address on the network) must be a multiple of the step factor. • What type of address is 148.76.21.0/20? • /20: step factor is 16 in the 3rd octet, but 21 is not a multiple of 16. Therefore it must be a usable IP address in this subnet • Identity address: host bits all 0's, multiple of step factor. 148.76.16.0 • Broadcast address: host bits all 1's, also the address below the next network. • Step factor is 16 in 3rd octet, so next network is 148.76.32.0 • So broadcast address on the 148.76.16.0 network is 148.76.31.255
VLSM: Allocating Addresses • Variable Length Subnet Masks: process of dividing an IP allocation into subnets of different sizes, with different masks • Requires CIDR routing to advertise the different sizes. • Several ways of doing VLSM. Here are my rules: • Work out the required size for each network. Leave room for the identity & broadcast address. • Allocate networks from the biggest down to the smallest. • After each allocation, leftover addresses. Use from the biggest remaining down to smallest for future allocations. • Keep networks of the same size adjacent in the numbering. Keep networks connected to the same router adjacent in the numbering, too.
VLSM: Other Rules of Thumb • If you split a /N range into two, you end up with two /N+1 ranges. • Split into 4, get /N+2 ranges. Split into 8, get /N+3 ranges etc. • Remember to watch the step factor as you are subdividing and allocating addresses! • Remember that links between 2 routers are also networks. Allocate a size /30 subnet to each one. • Place point-to-point router address allocations at the bottom of the original address range.
VLSM: An Example • We are given 193.85.72.0/24 to VLSM subdivide • Four LAN subnets. Two router-to-router (/30) subnets • LAN 1: 20 hosts. Closet power of 2 is 32. Use /27 • LAN 2: 12 hosts. Closest power of 2 is 16. Use /28 • LAN 3: 25 hosts. Same as LAN 1: /27 • LAN 4: 87 hosts. Closest power of 2 is 128. Use /25
VLSM: An Example • Start with 193.85.72.0, allocate the /25 (step of 128) • Next is 193.85.72.128, allocate one /27 (step of 32) • Next is 193.85.72.160, allocate next /27 (step of 32) • Next is 192.85.72.192, allocate the /28 (step of 16) • We have allocated IP ranges for the LANs • We are up to 192.85.72.208. From here to 255 not yet used
VLSM: An Example • We are up to 193.85.72.208, with only the two /30 router to router links left to allocate • We have 256-208= 48 IPs left, but only need to allocate two /30 subnets (4 each), so we still have lots of IPs left • Start at the bottom and allocate two /30s (step of 4) • 193.85.72.248 /30 • 193.85.72.252 /30
Route Aggregation • A set of routes can be aggregated or summarised when there is a single route with a smaller mask prefix which covers the set of routes exactly • This lessens the RAM requirements for every router, and also allows them to find the correct route for each packet more quickly • Special conditions to make this occur. It's not easy to develop a design to do this
Example Route Summarisation • Say we want to advertise a summarised route of 180.1.1.0 /25 • This is the range 180.1.1.0 to 180.1.1.127 (step factor of 128) • We can break this into two /26 subnets (size 64 each) • 0 to 63 for one of them, 64 to 127 for the other • Let's subdivide the second one into two /27 subnets (size 32 each) • 64 to 95 for the first, 96 to 127 for the second • We now have one /26 and two /27s which exactly cover the original route of /25. • So, if we had these routes: • 180.1.1.0/26, 180.1.1.64/27 and 180.1.1.96/27 • We could summarise them as a single route: 180.1.1.0 /25
RIP Version 2: CIDR Support • The first version of RIP does not support VLSM or CIDR • On Cisco routers, run show ip protocols to see what version of RIP your router uses to send advertisements • To switch to RIP version 2:
RIP Version 2: Summarisation • RIPv2 still summarises network on classful boundaries by default • To stop this from happening: no auto-summary
RIP Version 2: Static Routes • RIPv2 can advertise existing static routes too • Use the redistrubute static command
RIP Version 2: Debugging • RIP v2 debugging enabled: you can see the advertisements with the masks for each route
RIP Version 2: Common Issues • Not running RIP v2. Not enough network commands in config