Internetwork Packet Exchange and Sequence Packet Exchange. Basically, Novell's version of TCP/IP.

Novell designed NetWare from XNS technology, and designed the IPX/SPX protocols to eliminate the need to number individual network nodes. IPX and SPX reside at layer 3 and 4 of the OSI. IPX is connectionless, like IP; SPX is connection-oriented like TCP. On a LAN using NetWare, a network number is assigned for a given segment and MAC addresses are used in conjunction with this to generate their own station address. Though IPX is based on XNS, the latter uses standard Ethernet encapsulation. IPX packets before 4.x used a Novell-specific version of IEEE 802.3, although administrators typically removed the default in favor of Ethernet_II. Novell caught on and changed the default for 3.12 and 4.x to conform to the 802.2 standard.

SPX was derived from the XNS Sequenced Packet Protocol (SPP) and is typically used by a printing process or an application gateway. It adds more fields to the layer 3 header, which include source and destination connection IDs, and sequence and acknowledgment numbers. The 3.x default is to transmit keep-alives every 3 seconds. The 4.x default is every 6 seconds. These values can't be configured on the server side. On the client side, you can have it set for up to one keep-alive/hour. You can also implement keep-alive spoofing on Cisco routers that are between a client and a NetWare server. That's what the voices told me, anyway.

IPX and Cisco

The only proprietary part of Cisco's implementation of IPX is the use of IPX over X.25 or T1 connections (there must be Cisco routers at both ends). With Cisco, you can also integrate EIGRP into an IPX network running RIP and SAP. Their implementation provides all the functions of a Novell router and further capabilities for the IPX network:

  • Incremental SAP updates. By default, IPX RIP routers spew out RIP and SAP updates every 60 seconds. EIGRP sends out only changed information only when changes occur. When you configure this, you can retain RIP, which is better over bandwidth-sensitive connections, or turn off RIP, in which case EIGRP replaces RIP as the routing protocol.
  • The ability to increase the hop diameter limit from 15 to 224.
  • Optimal path selection. The RIP metric is based on ticks and hop count. IPX EIGRP uses a combination of delay, bandwidth, reliability, and load.

The end. Go home now.