Networking Technologies

Chapter 8: IPX Protocol Stack

 

 

Objectives:

Chapter 8 discusses the IPX/SPX protocol suite. The objectives important to this chapter are on page 8-1:

  1. Describe the purpose of Internetwork Packet Exchange (IPX) and Sequenced Packet Exchange (SPX).
  2. Identify and describe the components and characteristics of IPX addresses.
  3. Identify the components of the IPX/SPX protocol suite as they relate to the OSI model.
  4. Describe the IPX routing protocols RIP and SAP.
  5. Describe link state routing with IPX NLSP and compare RIP/SAP, NLSP, and OSPF functionality.
  6. Identify other IPX network services and what they do.
  7. Explain the benefits of protocol analyzers in managing an IPX/SPX network.
Concepts:

The IPX/SPX suite was developed by Novell, based on the Xerox Network Systems (XNS) suite. (Makes you wonder, did everyone who made money in computers "borrow" some ideas from Xerox?) IPX is connectionless, SPX is connection-oriented.

Page 8-3 lists four elements used in IPX addresses:

  • Network address - every node on a given network shares the same network address. This address is 8 Hexadecimal characters long (4 bytes). Addresses should be registered with Novell to guarantee that they are unique, if you will be making IPX connections to other networks.
  • Internal network address - only servers have IPX internal network addresses. Like network addresses, they are also 8 Hexadecimal characters.
  • Node address (also called MAC or NIC address) - every device that has a Network Interface Card has its own MAC address. It is 12 Hexadecimal digits long (6 bytes). IPX uses the actual MAC address assigned by the manufacturer as the node address for a device. This means that the node address will change when you replace a NIC. (The exception is ARCnet. ARCnet addresses are hand configured using DIP switches on the cards.)
  • Socket number - a socket number identifies a process or port or service or program that is running on a computer. Packets are addressed to communicate directly with specific services on servers and nodes. Some common socket numbers are:
    • 0451 = NetWare Core Protocol (NCP )
    • 0452 = Service Advertising Protocol (SAP)
    • 0453 = Routing Information Protocol (RIP)
    • 0455 = Novell NetBIOS
    • 9001 = NetWare Link Services Protocol (NLSP)

Every network that connects to another network must have a unique network number. Every node on a network must have a unique node number. Three numbers cannot be assigned to nodes, networks or servers: 00000000, FFFFFFFE, and FFFFFFFF.

If you examine the chart on page 8-7, you will see that the IPX/SPX suite, like the TCP/IP suite, does not address issues related to the ISO Physical Layer. This is for the same reason, that the suite is meant to be independent of Physical topologies. You may feel a bit better that the observations about the IPX/SPX suite are only related to the ISO model.

The IPX protocol is discussed in detail, starting on page 8-8. IPX performs network addressing and internetwork routing at the Network Layer. Routing can be based on RIP (Distance Vector) or NLSP (Link State) information.

The SPX protocol provides connection-oriented packet delivery at the Transport Layer. In this suite, IPX performs the delivery, but SPX makes sure the packets get there. The four Transport Layer topics are present: service addressing, address/name resolution, segment development and connection services.

The next section of the chapter discusses Routing Information Protocol (RIP) and Service Advertising Protocol (SAP). Be aware that IPX RIP is not the same protocol as IP RIP, even though they are similar.

RIP features are listed on the next several pages:

  • Routers that use RIP send RIP packets when they are turned on, and when changes are made
  • Entire routing tables are sent in RIP packets
  • By default, RIP packets are also sent once a minute, in addition to the event driven transmissions above
  • When routers are taken down, they broadcast a signal that they are now 16 hops away
  • RIP packets can be Requests or Responses
  • RIP keeps track of hops and ticks (1/18 of a second) to destination networks
  • RIP means that routers are constantly sending and receiving information that is secondhand or worse

SAP features:

  • SAP packets are sent in broadcasts, so they do not go beyond routers
  • SAP agents run on routers and keep track of services available on the networks those routers serve, storing the information in service tables
  • SAP agents broadcast information to each other so routers are aware of services on network segments that they do not directly serve
  • Periodic SAP Information Broadcasts are sent by servers each minute, by default, to advertise services
  • SAP Service Queries are used to find services
  • SAP Service Responses come in two types: General Service Responses are responses to broadcasts, Nearest Service Responses are responses to queries.

RIP and SAP are verbose protocols. They create a lot of network traffic. SAP traffic can be a problem across WAN links, so it is advised on page 8-17 that filtering out either incoming or outgoing SAP traffic is desirable.

A discussion of NetWare Link Services Protocol (NLSP) begins on page 8-18. This is the Link State protocol for IPX/SPX. It has the features you have been told to expect from a Link State protocol: uses less bandwidth, faster convergence and better scalability.

Some features of NLSP:

  • Specific terms: Adjacency refers to the state of a router being next to another router (one hop away); a Link is a connection between two adjacent routers; a Designated Router (DR) represents all the other routers on a network
  • NLSP routers send Hello packets to identify their adjacent routers
  • An Adjacency Database is a list of all the routers a given router is adjacent to
  • The Adjacency Database holds the following information about the Adjacent router:
    • System ID
    • Priority - an arbitrary value
    • MAC address
    • Holding time - a number of seconds. You expect to hear from the Adjacent router within this time interval, else it is assumed to be down. Default is 30 seconds.
  • The Designated Router for a network is the router with the highest Priority value. The administrator should assign a high Priority to a well placed, powerful router.
  • The pseudonode is a construct created by the DR. All routers on the network assume they are adjacent only to the pseudonode. The DR acts as the pseudonode. So the DR, in its secret identity of the pseudonode, receives all traffic from all routers and sends it on the the appropriate next hop.
  • A forwarding database is a table of routes. Routes are placed in a forwarding database if they are the lowest cost route to a destination, or if they are equal to the lowest cost route already in the database (providing choices). Cost is calculated, not just based on hops or ticks. Calculations take place after each hold down interval, every two hours by default.
  • The cost of a route is the sum of the costs of each link in that route. Link costs are assigned based on transmission speed:
    • FDDI (100 Mbps) = 14
    • Token Ring (16 Mbps) = 19
    • Ethernet (10 Mbps) = 20
    • Token Ring (4 Mbps) = 25
    • E1 (2.048 Mbps) = 26 (This is a European T-1)
    • T1 (1.544 Mbps) = 27
    • ISDN (US and European) = 45

Complete Sequence Number Packets (CSNPs) are sent about every 30 seconds. These are summaries of the links the DR knows about. Routers receive the CSNP and compare it to their link state databases. If the two match, the receiving router does nothing. If the CSNP has new information, the receiving router sends a Partial Sequence Number Packet (PSNP) to the DR, which is a request for the details of the new information. The DR receives the PSNP, and sends an LSP update on the new link to all the routers. If the CSNP had old information in it, the receiving router would have sent LSP updates to all routers.

The maximum number of hops for NLSP is 126. This is an improvement over the 15 maximum found in RIP.

Page 8-32 presents a comparison of NLSP and OSPF. They are very similar, but their terminologies are different.

NLSP - IPX/SPX OSPF - TCP/IP
Begins process Hello Packet Hello Packet
Update packet LSP LSU
Summary of database CSNP DDP
Request for update PSNP LSR
Acknowledgment (none) LSA
List of routers 1 hop away Adjacency Database Neighbors List
List of all routes in network Link State Database Link State Advertisement
Database
List of usable routes Forwarding Database Routing Table

The IPX/SPX suite also supports other NetWare protocols and features:

  • NetWare Core Protocols (NCP ) are the basic network services in NetWare
  • Multiple Link Interface Driver (MLID) is a network board driver that can be bound to multiple protocols
  • Link Support Layer (LSL) is a switchboard-like interface between MLIDs and upper layer protocols
  • Application services provided through NLMs

The chapter ends with a discussion of protocol analyzers. A bit more detail is offered, compared to the last chapter. You should know that:

  • A Time Domain Reflectometer (TDR) is used to analyze Physical layer problems, like cable breaks.
  • Data-Link Layer analyzers look at traffic and packet problems.
  • Network Layer analyzers look for client-server and router problems.
  • Upper Layer analyzers look for problems in protocols from the Transport Layer on up.

A list of analysis tools is offered: the TDR, three variations of LANalyzer (including ManageWise), and Sniffer (from Network General).