The purpose of this paper is to explain how IBM may choose to adapt SNA to accommodate the presence of OSI. It defines three different approaches available to IBM: adapting SNA to become closer to OSI; providing network gateways between SNA and OSI; and running SNA over the OSI lower layers. It then discusses the pros and cons of each approach. Interestingly, the authors do not discuss converting IBM products to use OSI protocols as a possible approach.
In many respects this paper reads more like a business magazine survey than technical paper. The technical content is often superficial and relegated to tables and diagrams that are not fully explained in the text.
The paper also has some technical errors. The authors state that routers are “protocol-dependent” and “therefore one router is needed for each protocol used by the network.” This statement is not correct. Routers can support multiple distinct network layer protocols; they are protocol-dependent in that routers can only route network layer protocols they support, unlike bridges, which route at the MAC layer and thus can route packets for any network layer protocol. Another problem appears in Table 1, which states that “network management is difficult in open systems.” The widespread deployment of SNMP in TCP/IP systems is a clear counterexample to this assertion. Finally, the authors use the term “internet transmission” to refer to situations when one protocol suite is encapsulated in another protocol suite. The choice of the term “internet,” which has connotations of internetworking (in which a common network layer protocol is used to interconnect diverse networks) is unfortunate.