Saturday, March 27, 2004

Dinosaurs and Mammals

Large software corporations like Microsoft and Oracle in addition to groups of software developers like elements in the Open Source community are giving the world new versions of the same thing: monolithic software for a world that no longer needs or wants such products.

Take for example Microsoft Exchange Server: this is a heavy-duty, if often ridiculed, system of software. This software was meant to handle hundreds to thousands of users and their information. What about the company that wants a Microsoft solution but only has 50 users? They have to pay for Exchange. Yeah, you can get a great discount, etc., but I'm not talking about price -- I'm talking about need. And designing software for need.

Oracle is more guilty of this sin. Let's say I need to have my data highly distributed on remote machines, but that I don't have tons of data and I don't need the horsepower Oracle can provide. Yet, if I use their product or any other large RDBMS like PostgreSQL, SQL Server, or even MySQL, I am using the wrong tool for the job.

Another example: Apache is the world's most popular Web server, but let me commit heresy and ask "Why?" How many people need it's speed? Or its capacity to handle millions of requests? Think of the hundreds of thousands of web sites out there that are visited once a day or once a month. Using something like apache for these is engineering overkill.

Microsoft and Oracle as well as organizations like the Apache Foundation, MySQL, and PostgreSQL are producing software dinosaurs. The have lived, they have thrived, they have helped define our information universe. That universe is changing. With the introduction of programming languages and development processes that foster rapid development, for the first time in the history of modern technology, we are in a place to design enterprise software around the needs of the end user (individulas and organizations), not around the needs of licensing structures and coporate philosophies. Imagine that the motivation behind software development wasn't market domination, but rather market diversity. The right tool for the right job.

I haven't counted the lines of code in the apache web server, but let's just say it's 100,000 lines. What if I only need to serve a couple pages per second and a max of 100 concurrent sessions? What if I could use Web server software that could provide what I need in only 200 lines of code? 100? 6? Why should I use Apache for projects that don't require Apache's strengths? The same thing for my email server and calendar application. What if I could do the same thing for all my networked applications? What if in addition to being small, this software was highly distributable as well? What if I could build office applications perfectly suited for a firm of 40 employees, and they didn't have to pay for licensing of the cliched software products? In addition, as the firm grew, this software could support 100, 1,000, or 100,000 users just as easily simply by adding the required light-weight components.

Software like this is easier to setup and maintain. It's easier to get involved with and understand. It's easier to modify, add to, customize, and enhance. It's easier to provide your firm with what they need.
Enter the new software and application engineering model: the mammal. This is where we are heading. Imgine application servers so light-weight they could run on a 9-volt battery for a week. Imagine them embedded in hundreds of devices all over the country or even the globe... so cheap and distributed, it didn't matter if one went down or even if twenty of them went down. They are quick, nimble mammals, not lumbering dinosaurs.

There are already software developers working on projects with views and goals along these lines. Watch for them: they are small and furry. They move quickly and quietly. Pay attention to them; they will one day rule the world.


No comments: