Thursday, June 09, 2005

My Thoughts on Portals

I work for a company that makes its fortune off of building portals for companies using IBM Websphere Portal Server. Personally, I've only been involved with one portal project (thankfully), but it gave me some real insight into why portals kind of suck.

First of all, I cannot believe how much money companies are willing to pay for what little they receive. I mean, just the price of the software alone is outrageous, and that only gives you a basic framework and a sample portal. I hope you didn't want anything relevant to your company for that $50-100k, because that will cost you extra! But, hey, it's IBM software written completely in Java (meaning that it's not Microsoft, so it's worth the price, right?)

So, let's say that a company just dropped $50K (or whatever the going price is) to buy Websphere Portal Server Enabled. Who's going to install and configure it? The local IT department???! Really smart people from my company who have been working with Websphere for years still have trouble getting all parts of the thing up and running, and keeping it running smoothly. And each new version of the product introduces new “bugs” that make it impossible to guarantee a perfect installation by a knowledgeable person.

Oh, and I hope you have a lot of new/powerful hardware with LOADS of memory and disk space (because remember, IBM is a hardware company first!), because it ain't gonna run for you unless you do. Our development environments require 1GB minimum RAM just to be usable. Plus, you'll want to cluster your production environment across multiple servers to improve overall performance, which introduces a whole heap-load of complexity to the configuration.

I'll have to say, though, that once a WPS portal is completed, that the whole thing is quite impressive. Perhaps in the same way that it is an awe-inspiring feat of engineering that we can send three people to the Moon and back, but in the end, you think about just how much money it took and wonder if the ROI was ever really there.

I'm not going to give a free pass to Microsoft SharePoint, though. While I'm happier with what they're calling a Portal with version 2 (versus the first version, where they slapped the Digital Dashboard onto the document management system and called it a portal), overall, it falls short of where it needs to be in order to draw companies away from WPS. I'll probably detail those shortcomings some other time when I have a comprehensive list, but my point is that SPS does not yet directly compete in the same space as WPS. And that gives my company a huge reason to continue pushing J2EE over .NET.

In my naive opinion, portal platforms are missing a key opportunity by making the technology server-centric. I mean, sure, being server-centric allows anyone from anywhere to connect to the portal with whatever browser is available on the machine that they happen to have access to. But, it also sends us back to the mentality of the mainframe and hub-spoke infrastructure model where you have an all-powerful machine in the center doing all of the work while dumb terminals exist only to provide the user an interface to the brains of the system.

My vision of portal technology places the aggregator at the client instead of the server. Or at least the concept of a “portal server” would be closely tied to one user instead of all of the users of a system. Let this “Personal Portal Server” be responsible for aggregating content and generating/caching the presentation. Then let the user decide what information needs to be displayed, even if that means combining Portlet/WebPart information from multiple providers onto one page.

If Google Maps and GMail has taught us anything, it's that the client (even a web browser) can offload a lot of processing from the server, and this is not much different than what I'm suggesting here. Centralized servers do not completely go away, but become specialized for the task at hand—providing specific content to a whole host of remote aggregators.

This concept really needs some refinement to become practical, but this is just a View Inside My Head at this moment.