The Best (and Worst) of 1997
By all measures, 1997 was a very good year for the computer networking industry. New and exciting technologies gave birth to strong products, which in turn helped many a bottom line. This says nothing about the benefits these products passed on to the network managers who actually bought them and did the hard part: proving that the products offered some sort of use beyond grist for the marketing mills. Indeed, there were many winners this year.
There were also a fair number of loser technologies, products, and companies, as well as the as-yet-to-be-decided contenders who offer strong possibilities, but which have failed to execute in one way or another.
Without a doubt, the most important development in the industry this year
has been the rapid and wide-spread adoption of the Internet
Message Access Protocol. Although no one vendor has yet to offer an IMAP
server or client that provides all of the functionality that I need, the technology
has matured to the point where I can choose from several
I use Netscape Communicator's Messenger component as my IMAP client because of its support for connected and disconnected modes, and because it works with my mail server (always important!). I also use Pine on my LINUX host when I need to Telnet into the mail server, although this is strictly a secondary option. I tried using Microsoft's Outlook Express, but it couldn't locate my server-based folders. I've never been a fan of Eudora, and have not yet tried their new implementation.
For my IMAP mail server, I use the University of Washington implementation, mainly because I could get it to compile on my LINUX host. It also integrates directly into sendmail, allowing me to use it along with the hundreds of other sendmail-based add-ons, plus it comes with POP daemons that provide legacy access.
2160: There's been a lot of buzz about PCS phones in the past
year or so. However, most of these deals (like PacBell's PCS service) are
purely digital. Pure-digital PCS sucks for the most part, due to the lousy
signal quality and low coverage area. For full-time usage, analog AMPS still
rules the roost. The great thing about the 2160 is
that it's dual-mode, supporting both the broad-based TDMA digital network
as well as simultaneous access to the analog AMPS network.
This lets me use AMPS for voice while simultaneously using digital services such as encrypted dialing, caller ID, short-text messages, integrated paging, voice mail indicators and the like. Another example: I can use my laptop's modem with the analog circuit, and still receive pages over the digital circuit at the same time! It is the best of both worlds in one handy unit. Although other vendors offer the same sort of dual-mode access, only Nokia has the 3-line alphanumeric display, critical for use as a paging device. There are also a large number of after-market products available, from hands-free kits to cellular-aware modems, rounding out the total product nicely.
Although the 2160 was released in 1996, the product has only begun to live up to its full potential this year. This is Cellular One's fault though, as the phone has been a bitchin' little multi-purpose communications unit since its release. Unfortunately, you'll have to use Cellular One if you live in the Bay Area and want to use this phone. The national PCS dealers are, as I said, strictly digital for the most part. Although the 2190 is a PCS-only equivalent of the 2160—and still a damn fine phone —the dual-mode nature of the 2160 is what makes it such a kick-ass gizmo.
- Intel NetportExpress:
When I reviewed this product for InfoWorld back in October of '97, I gave
it a less-than-perfect score. But after having lived it with for a while,
I'm now convinced that the NetportExpress is the best standalone print server
on the market, and another of 1997's seminal releases.
The product offers what is perhaps the broadest range of integrated protocols, ranging from AppleTalk to IPX, along with other neat features like automatic conversion of LF to CR-LF, and even FTP uploads to a printer port. This latest feature is pretty neat; you can use any old FTP client to send files to a printer! It also has the most full-featured web-based management of any print server that I've seen. Anybody in the market for a print server should look at this one. A "Buy It Now" kinda product. Intel, I'm sorry for not giving it an "Excellent" in the review.
- NetXRay 3.0: Cinco Networks developed this killer software-based network analyzer a couple of years back, and has turned it into one of the best overall analyzers on the market. So good in fact that the company was bought out by Network General earlier this year. Running on Windows NT and Windows 95, version 3.0 supports things like graphical protocol distribution maps, real-time protocol decodes, support for Network General's Sniffer file formats, and a host of other features too numerous to mention. Take my word for it, this is one of the best damned products released in 1997; I use it on a weekly basis. Another "Buy It Now" product.
I've become addicted to WRQ's Reflection Suite for TCP. It's got everything I need from a TCP/IP applications suite, and follows most specs right down to the letter. Most of the other suites are a joke by comparison, although all have their stronger points in one area or another. However, WRQ's excellent support and overall good-naturedness as a company make them a pleasure for any firm to partner with.
Finally, Funk Software's Steel-Belted RADIUS is one of those products that you never think about once you've got it. But if you don't have it, you wish you did, and on a daily basis. For anybody doing RADIUS authentication on dial-in servers, this is the only thing you need. The only product I've ever reviewed that I've given an "Excellent" rating, by the way.
- Apple: Proving that
Apple really is an arrogant lot, the death
of the Apple clone program marks the end of the company's mainstream market
presence. Steve, rather than asking me to help Apple by buying a Mac,
why don't you give the market what it wants: a consumer-focused strategy that
gets your only permanent technical asset (the MacOS) into the hands of the
masses. Vertical markets are for entrepreneurs, not established brands.
I've got an old IIvx here that I love, but unfortunately can't get much software for since it's not a PowerPC-based system. I was going to buy one of the Power Computing or Motorola clones, but now I'm going to wait-and-see what happens with Rhapsody instead, just like the rest of the mainstream market. If it runs fine on Intel, then I won't be buying an over-priced, end-of-the-line Mac from Apple, anyway. Stupid is as stupid does.
Although the Dynamic Host
Configuration Protocol has taken off as a mainstream technology, the weakness
of the implementations are making the protocol in general suffer. Every little
router and micro web server is coming with a bundled DHCP server, but they
won't honor specific requests, or they respond to any request without checking
the assignments first, etc. Even the RAS server in NT, which can use DHCP
to allocate addresses to dial-in devices, doesn't
pass on the default gateway, DNS servers, or other information readily available.
Hey engineers! How about following the @#!% specs!
Meanwhile, the good providers of DHCP aren't getting leveraged, either. Part of DHCP's beauty is that it lets you define things like the default mail server, the browser's homepage, and other application-level settings. As far as I am aware however, not a single mainstream TCP/IP application takes advantage of these already-existent options. This means administrators still have to define these settings at every desktop across the organization, revisiting them whenever the Intranet's URL is changed. While ACAP promises to address these problems and is building momentum within the industry, we've already got DHCP servers in place that already do the same thing, albeit on a lower scale. This is just another example of DHCP losing merit.
- Microsoft Office 97: Incompatible file formats drive the Microsoft revenue engine. I refuse to play the game any longer, and it looks like I've got plenty of company.
- 56k Modems: 'nuff said.
- xDSL and Cable Modems: Sure, POTS is slow and ISDN is
expensive (at least in PacBell territory),
but I don't see xDSL or cable modems solving these problems very well.
A couple of reasons why xDSL may not succeed: it takes forever for telcos to roll-out the technologies they like, nevermind the ones they're unsure of. As we've already seen with ISDN, switch-based services tie up the telco resources in a big way, making it difficult for them to justify the rollout without major financial incentives (meaning you and me). ISDN had lots of promise, but when the telcos found out how much money they were losing it became a lower priority. Yes, there are lots of varietal implementations of basic DSL, but that only muddies the water even more. Consumers won't put up with mixed messages.
Cable modems? HA! Where's the infrastructure? I live in San Mateo, right in the heart of the most technologically-advanced community on the planet, yet I can't get cable modem service for at least a year, probably longer. TCI is the carrier of choice by city sanction, and they don't see the cost justification for swapping out their head-end equipment and stringing new HFC cable in my neighborhood. San Mateo isn't even on the list of planned cities. I can't blame 'em, and that's why I don't think its going to succeed. By the time it's prevalent, there will be other technologies that use the existing infrastructure sufficiently well to make the need for new ones irrelevant.
POTS and ISDN are here to stay people. xDSL and cable modems might work out in some areas, but I'm not holding my breath. We'll just have to see how these things shape up.
- NDPS: Novell has reinvented network printing with Novell Distributed Print Services, and in so doing has managed to turn a simple network service into a new job category. NDPS is as beneficial and complicated as NDS although after several years of best-effort trying, the latter still hasn't hit majority status. NDPS will likely take just as long to succeed just as poorly. However, as a technology, NDPS is a major boon to enterprise-class corporations, and if the market is built appropriately, NDPS could succeed. Only time will tell how this one unfolds.
- WebNFS: Whatever happened with this little gem? According to statements made this time last year, we were all supposed to be accessing network filesystems from within the safety of our web browsers by now. As far as I'm aware, nobody is doing diddly with WebNFS, either on the client or the server sides. Shall I move this into the losers category? Or will it even be relevant next year? Still, Sun could make this fly if they wanted to, and with the help of Netscape and others, the technology could live up to it's promise. We'll have to see if this little orphan can eke out its own existence, or if it dies of neglect.
These recommendations are a result of my own usage and observations. Some of these products are going to be totally meaningless in your environments, and rightly so. Likewise, many of the coolest products are going to be those that I haven't seen.