henry@utzoo.uucp (Henry Spencer) (03/21/89)
In article <1989Mar20.175543.4694@tmsoft.uucp> woods@tmsoft.UUCP (Greg Woods) writes: >>Topic: I've been confused by much of what I've been reading about >>'migration from TCP/IP to ISO'. >> >>Is this necessary? > >YES!!! I think it will make things simpler... No it won't. Ever read an ISO spec and compared it to the TCP/IP spec? The migration is necessary for political, not technical reasons. The ISO protocols, at least the ones seen so far, have *no* technical advantages over TCP/IP that matter. What's more, they are largely-untried protocols designed by large committees (whereas TCP/IP are thoroughly shaken-down protocols designed by a few people). They have a lot of political momentum behind them and are hence inevitable, unfortunately. The one bright spot is that *both* TCP/IP and ISO are likely to be obsolete within a decade anyway, since neither one appears to be capable of the performance needed for really high-speed networks. A cynic might suggest simply postponing ISO conversion until it's clearly obsolete and not worth bothering with -- everybody has to be able to speak TCP/IP for now anyhow. >>Is it safe for a new network to just start with ISO now? > >Do you have any machines using TCP? If not, I think the answer is >yes... That depends on how much you value the many man-years of experience with TCP/IP. Pioneering new protocols tends to be painful. *I* wouldn't call it "safe" for a production system. Also, are you sure you'll *never* have any TCP/IP-only machines on your network? Even the Corporation for Open Systems, which is dedicated to promoting ISO, found it had to use TCP/IP internally for a while. -- Welcome to Mars! Your | Henry Spencer at U of Toronto Zoology passport and visa, comrade? | uunet!attcan!utzoo!henry henry@zoo.toronto.edu