DTJ Number 3 September 1987 - Digital Technical Journals

Editor's IntroductionRichard W. BeaneEditorThis third issue features papers about Digital's networkingproducts. Digital was an early advocate ofdistributed interactive computing, a conceptallowing systems resources to be shared amongmany users over a network. Just as two personsfrorri different cultures have problems communi·eating, however, so do computers with differentdesigns. Some standard set of rules is needed toallow successful interaction.The Digital Network Architecture (DNA) is theset of rules that defines how Digital's productscommunicate over a network. Being flexible, thisarchitecture allows many ways for design ·groups toimplement the DNA rules into various DECnetproducts.The first paper discusses the DNA structure andhow it has evolved. Tony Lauck, Dave Oran, andRadia Perlman describe DNA's design goals and thenew functions supported in its four developmentphases. The tasks performed by the eight DNA layersare explained, with particular emphasis on thenetwork management and routing layers.To achieve high performance, models and simulationswere used to test the DNA structure. Thepaper by Raj Jain and Bill Hawe relates some casestudies, one for each layer, that resulted in fastercommunication. These models helped to optimizehow data packets.are handled by simulating differenttraffic patterns.Although the DNA and SNA architectures arequite different, they can communicate through theDECnetjSNA Gateway product. John Morency,Dave Porter, Richard Pitkin, and Dave Orandescribe how the gateway's design accomplishesthis communication. The authors describe thecomponents in each architecture and how messagesare structured.The paper by Bill Hawe, Mark Kempf, and AIKirby reports how studies of potential new broadbandproducts led to the development of theExtended LAN Architecture. The design of theLANBridge 100, the first product incorporatingthat architecture, is described, along with thetrade-offs made to achieve high performance.The speed of communication between terminalsand systems depends on how they are connected.Bruce Mann, Colin Strutt, and Mark Kempf explainhow they developed the LAT protocol to connectterminals to hosts on an Ethernet. The EthernetTerminal Server, the DECserver 100, and theDECserver 200 all use this new protocol.The next three papers describe how DNA wasincorporated into three different operating systems.The first paper, by Paul Beck and Jim Krycka,explains how the DNA principles were built intothe VAXJVMS system. The authors describe howtransparency was achieved by a tight couplingbetween the VMS software and the DECnet structure.The ULTRIX software is Digital's second operatingsystem for its VAX computers. In the secondpaper, John Forecast, Jim Jackson, and JeffSchriesheim describe how they blended DNA intothe ULTRIX software. Several unique tools weredeveloped to avoid changes to existing DNA implementations.The DNA architecture has also beenincorporated into the MS-DOS system in theDECnet-DOS product. The third paper, by PeterMierswa, Dave Mitton, and Marty Spence, describeshow they built communication services intoMS-DOS's background by writing new code andborrowing existing code from the DECnet-ULTRIXsoftware.The final two papers discuss an important aspectof any network: its management. Nancy La Pelle,Mark Seger, and Mark Sylor discuss how networkmanagement is built into many diverse DECnetproducts. They describe Digital's common managementarchitecture and the need to meld themanagement of voice and data networks. TheNMCCJDECnet Monitor controls a DECnet networkfrom a central location. Mark Sylor relates how thismonitor functions, describing its database structureand reports for the network manager. Themonitor's analysis techniques to identify real-timeproblems are especially interesting.I thank John Adams, Andrea Finger, and WaltRonsicki for their help in preparing this issue._2) 02