Networking's Think Thin!
Well, here it is! The end of 1999 and all of the stories, concerns and speculation about the Y2K "bug" will soon be unveiled. Soon, we'll be able to get back to business and work toward improving our companies and bettering business processes.
2000 should be an interesting year for those of us in Information Systems. For one, what will we talk about now that Y2K is a memory? Other items of interest will include where the AS/400 goes from here. IBM's marketing makes the AS/400 look more and more like an also-ran every day. Can the AS/400 survive an attack from within its own company?
Client/Server computing, networking, Java, and thin-client computing will continue to evolve and become more robust and hopefully simpler to work with. Networking in general has evolved at a constant pace since the mid-eighties. On the hardware side, hubs have gotten faster and switches are becoming more prevalent and replacing hubs, providing even faster access. Still, the basic technologies don't seem to be changing much. Ethernet is still and will likely remain the clearly dominant topology. TCP/IP is due to be expanded since the existing address scheme (i.e., 255.255.255.255) is running out of numbers.
Be prepared. The migration to the next level of TCP/IP addressing will either be a complete non-event and extraordinarily easy or it will be a nightmare. Which one it is depends on who designs the changes and how practical and averse they are to generating billions of dollars of unnecessary IT infrastructure costs. It seems to me that the fix to this problem could be as simple as prefixing 000.000 to the numbers already out there and then moving forward with 000.001. But I'm sure there are many people smarter than I who will fire off a litany of reasons why this cannot be done this way.
If your network is not yet pure IP, if you are running multiple protocols, or if you don't have the ability to connect your network to the Internet, be prepared to look into these in the New Year. Networks running multiple protocols tend to be slower, less stable and more difficult to troubleshoot, and as a result are more expensive and provide no real value in return.
Will 2000 be when thin is finally in? Well, maybe not completely in, but as Java continues to evolve and become more robust, Web-based intranet and Internet applications will become more prevalent. Those companies that have done or tried to do large scale, fat-client deployments of complex client/server applications across a WAN have learned that this is a task not to be taken lightly. It is extremely difficult and requires a well-oiled machine consisting of a well-balanced high-speed network, excellent software deployment methodologies and DBA skills, and most of all a high caliber IT team that communicates.
To be honest, I don't believe that I've seen that combination in practice yet. As a result, two technologies will continue to grow in popularity in 2000. The first, Java, has the benefit of general appeal and low licensing costs along with thin-client deployment. The second, Windows NT Terminal Server Edition, preferably used in concert with Citrix MetaFrame, is also a popular thin-client solution. The drawback here is that the licensing costs can be steep. The good news is that network bandwidth requirements are minimal.
Look forward to continued improvements in networking next year, but keep in mind that if you're in a Wide Area Network and doing client/server, thin is in!