The Best (and Worst) of 1998
1997 was a banner year for the networking industry, bringing a hoard of new technologies and products to the newly-networked masses. 1998 was the morning after, with most of us trying to make 1997's technology work, while vendors pushed out follow-on products that tried to get closer to the original promise. Sometimes the fixes worked, with some products and technologies permanently altering the landscape, while others just proved that some technologies weren't really such great ideas in the first place.
Here then is my short list of the winners and losers for 1998.
- Linux on the Back-End: Although I've been writing
about Linux on the back-end for over a year now, the rest of the world
didn't pay much attention until Oracle and Sybase joined the party. This
acceptance by leading back-end vendors—combined with a
monster NOS platform that is the best of the lot—has resulted
a whopping 211
percent increase in shipments for Linux servers.
We're still missing key components like enterprise-class application servers (I'm still waiting for my SuiteSpot beta), storage-management solutions (somebody wake up Legato), middleware (Allaire?) and so forth, but if nothing else, the platform's acceptance has proven to be the network-technology story of 1998.
- Sonic System's SonicWall: If you've got a permanent Internet
connection, you need a firewall. It's really that simple. For small networks,
the best solution on the market is the SonicWall
firewall from Sonic Systems. It's a self-contained, hardware-based stateful-inspection
firewall, VPN bridge, NAT and DHCP server all rolled into one very inexpensive
and easy-to-use product. Better yet, the company provides frequent and free
updates to their firmware, and also offers exceptional support.
Every freaking day some twit sends oversized packets to my web server, runs port scans on my mail server or TCP probes to everything on my network. The SonicWall stops them all, yet allows me to configure whatever connectivity options I want without having to setup proxies or gateways or any other crap. If Linux is the year's killer technology, then this is the year's killer product. There are still some features missing (like integrated authentication and detailed reporting), but its better than software-based firewalls any day, and is only a fraction of the cost of competitive hardware solutions. It's the network security solution for the rest of us. (Note: I wrote a review of this product a while back, but Sonic fixed every complaint I had. It's a killer now.)
- Novatel Wireless' Minstrel: My second favorite product
of the year is a CDPD
modem released by Novatel Wireless for the PalmPilot, although my love
is more for what the product lets me do rather than what it does by itself.
This unit provides CDPD-based Internet access to the PalmPilot, letting me
use Actual Software's MultiMail to
check my IMAP server and Oku Kazuho's Palmscape web
browser to access my on-line applications. I don't need no stinking phone
lines, laptops, or goofy markup languages: I
can get to any Internet-based or corporate data I want using a gizmo the
size of a cheap romance novel.
Yeah, CDPD coverage still sucks, but this is a monster combination of extreme portability, standards-based content, and wireless connectivity. It's a must-have for any serious road-warrior who doesn't need a portable desktop all of the time. I'd love to see something similar for NAMPS. Anybody?
- Java: Between Sun's screwy licensing demands and the development community's issues with portability, stability and performance, the fate of Java is in serious question. But beyond that, it's already seen a tremendous number of failures. The early part of the year saw such spectacular abortions as WordPerfect and Javagator, and now we're seeing that it won't hold up as a back-end server platform, either. It just can't meet the performance and stability levels required of a database server, a virus-checker, or a tape backup engine. All that's left for Java is where it first started: a cross-platform run-time engine for simple applets. It's competition isn't Windows, it's Visual Basic.
- NCs: All we wanted were dumb terminals with an embedded web browser. What the MBAs gave us instead were stripped-down PCs running Windows applications, 3270 emulation and anything else that could fit onto a hard disk. Once the PC makers got their costs down to $500, they beat the NC vendors at their own game, giving more functionality at the same price. I still say there's a market for a dumb terminal running Navigator (and only Navigator) in firmware. Too bad we can't buy it from anybody. Vendors say the market never materialized. I say vendors never delivered on a product that made sense.
- NetXRay: Last year, I said that NetXRay 3.0 was one of the best products released in 1997. Cinco took a great core technology and made it easy-to-use and affordable, expanding the network monitoring market at the expense of the high-end players. Things looked great for us. Boy was I wrong. Network Associates has since quadrupled the price, totally alienating the entry-level market that Cinco had built. When that didn't kill the product, they went ahead and broke it, removing some of the very functionality that made it awesome to begin with (like free application-level decodes and real-time monitoring). What a bunch of dorks. Now I tell people to get Shomiti's Surveyor, a bitchin' decoder that's a fraction of the cost, or Ethereal, a free (but feature-poor) decoder for Linux.
- Laptop Buyers: As the rest of the PC industry goes after the larger markets that affordable functionality provides, the laptop sector is stuck in navel-gazing mode. Tell me again why I need DVD video on my laptop? I don't know about the rest of you, but I don't equip my users with laptops so that they can watch movies. Worse still, this infatuation with pointless features is keeping us from getting stuff like embedded Ethernet cards, on-board floppy/CD-ROM combos and full-featured docking stations. Only the high-end models even have port replicators anymore. This whole sector ought to be taken out back and put out of my misery. Start over.
- Linux on the Desktop: This is a tough call. On the one
hand, Linux as an OS for everyday end-users has a lot going for it. Administrators
can setup a base install that only has a DHCP requester and some basic services,
loading everything else off a server-based installation. And the market for
productivity software is more active here than any other platform, with offerings
available from WordPefect, Star
Division, Applix and
But on the other hand, configuring PCMCIA support and other basic services is still a huge hassle, and there's lots of productivity applications that are still missing (hey, whatever happened to FoxPro for UNIX anyway?). The Linux community's unwillingness to unify behind KDE—a killer GUI for Linux that is easily equal to Windows or the MacOS—is also keeping this sector from moving forward. A lack of commercially-available (and supported) device drivers isn't helping anything, either. Add it all up, and there's still a long way to go.
- Broadband Access: This was on last year's list too, and I'm still undecided about the prospects. Although I've gotten ADSL access now, I'm still not sure the telcos know what they're doing with it. My line goes down for a couple of minutes every single day, for example. Its great when it works, but if PacBell can't even keep my line up, what are they going to do with a few million more users? Even when it works, I find that I've got more bandwidth than the sites that I'm getting data from anyway. And cable modems still don't have the infrastructure to support the load, and the shared-access nature of the medium makes it unsuitable for business use anyway. Maybe next year.
- XML: Too complex for your average HTML-coder, but too weak to be useful for complex data models. I personally don't expect XML to take off for everyday use, but it'll still show up in lots of products. I think we really need an Internet-standard protocol for exchanging data specifically, rather than yet-another presentation language. XML solves the wrong part of the problem, and does it poorly to boot.
- Voice over IP: Depending on where you sit, VoIP is either the Internet's next killer app or a boondoggle waiting to happen. For basic functionality, VoIP works just fine, with calls working pretty much across vendor lines. But press the "hold" button and you're likely to lose the call, H.323 is Internet-hostile in a number of ways, and it costs more to build a VoIP infrastructure than it would to do the same with a regular PBX. These are big problems. The promise of VoIP is that packetized circuits offer more functionality than what's offered by the "flash hook" on traditional circuits, but until vendors can get basic interoperability and costs under control, nobody will care. Good luck to us all.
I've Shown You Mine, Now You Show Me Yours
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.