02 January 2011
I can't believe how quickly things have come and gone. I thought 2009 was great, but 2010 year no doubt takes the cake! Now that we're all on the other side of the new year, I thought I'd take a quick look at my year (and my decade) and talk about the Spring framework.
Last decade started strangely for a lot of us here in the United States. We saw a very strange election, the source of much discord in the political world. Scandalous though that might have seemed, it would all seem so fleeting when two planes crashed into the twin towers in New York City, New York. The reactions of a panicked country seemingly under siege would change the landscape over the next several years. The ramifications are still very pronounced and felt, I fear, in every corner of this tiny little planet.
I started the decade, and my professional career, wading through the dot-com boom and subsequent bust. I've been fortunate to have worked with a lot of startups over the years. I've also worked in stodgy, established corporations. I feel like working at a bank is the easiest way to fall through the cracks in life. It never ceases to amaze me that there are projects in some banks being started this very day, January 1, 2011 (1/1/11..!) using Struts 1.
It has been my experience that startups have the blessing - and burden - of having no room for failure. I have spent a lot of time making things work that just shouldn't work. While I've always been best-practices-oriented, I was leery of any gospel according to the vendors. I once had a manager utterly terrorize me because I was using EJBs when he had PHP script kiddies that purported to be able to deliver a huge function of the project done by "just downloading a script!" This taught me a valuable lesson. Looking back, I suspect the manager's claim was probably genuine, and I feel like such a heel for blindly relying on the vendors' claims that EJBs were the way forward, the future. I never could convince myself that the PHP way was the way, but there was no denying the obvious productivity gains for many common tasks. There was a time when I was looking pretty seriously at moving to .NET because it promised radically improved turnaround times. One of my main reservations about the then-nascent .NET ecosystem was the lack of open-source. I've always worked hard to use open-source if I could.
The Spring framework came along just in time. Engineers that survive in startups tend to be scrappy engineers who are interested in results, not process, and who aren't in it for the 9-5 shift, but for results. The Spring framework is open source; it closed the gap between .NET and Java by providing niceties like declarative remoting, declarative transactions, etc; it also helps further best-practices, not hinder them. I've been a true believer in the Spring framework from day one.
In early 2005, I moved to Tempe, Arizona to help start a company. The venture was a victim of the wayward economy when the housing bust hit. While I made some great friends, I was glad to return to Los Angeles in March of 2009.
Early in 2010 I had the fantastic good fortune to join a company here in Los Angeles called Shopzilla - they're the group behind BizRate.com, BeSo.com, Shopzilla.com, and tada.com, each of which gets a staggering amount of traffic. They use the Spring framework extensively and feature an excellent engineering environment as well as one of the best workplace environments in the business (arcade, lounge, free lunch, great perks, gym, etc., etc.). There are many brilliant people at Shopzilla. I think it is our natural tendency to approach strange codebases with trepidation and a sense of, "well, I could do this better!" and a lot of times, unfortunately, that instinct is on the money. I never felt that way with the large swaths of code I encountered in Shopzilla's codebase.
I've been in the unfortunate position in my life of having to work at places where engineers were begrudged unit tests; where code and projects were wildly inconsistent, or antiquated across projects; where there was a strong, undue sense of code ownership and code was walled off to all but two engineers (who had twelve weeks off yearly and were busy planning their retirements). Shopzilla was the antithesis. It was... wonderful to be completely unencumbered by management and bureaucrats.
I loved Shopzilla's earnest maximization of the Spring framework.
Let me try to explain. If you'll permit a generalization, engineers at startups can't afford to trust, anything. No time, after all. You hope that people already appreciate the benefits of code reuse, of course. I wonder if those people also realize the responsibility that comes with reuse. If something breaks in that code, its up to you to see it gets fixed. Keeping a stack as open-source centric as possible facilitates fixes. Broken code in a server is bad, but try proving it, let alone fixing it, without the source code! Support contracts might help there, but rarely as quickly as you need it to. It takes courage to use other people's code, and it takes courage to trust and verify promises extended by that code. It takes skill to integrate - and to care - about the big picture system, and not just the individual moving parts, too.
That is my conceptual framework when approaching projects. I use the Spring framework because it integrates with many different technologies and insulates users from the idiosyncrasies between various technologies through a unified component model. It levels the playing field. Depending on the stack, there are any number of places one might have to tailor a project 's behavior to the requirements of a system. Logging frameworks, service tiers, web service tiers, caching frameworks, web (including core APIs like Servlets), workflow engines, integration engines, processing grids, batch processing frameworks, job scheduling frameworks, etc., all have their own notion of a "component" in that domain, and each one of them expects you to work in terms of that component model, ignoring the uniformity and conventions you'd hoped to establish across your code base.
Spring opens the doors to all these technologies (and untold many others) and lets you work in terms of your domain model. The engineers behind the Spring framework have plumbed the depths of these various frameworks and done the legwork to make it easy for you to integrate them in a manner consistent with your codebase. Where they've had to, they've fixed parts of these projects, and provided less risky paths for integration. Often, this process is as simple as dismantling the custom component model in place and letting people think in terms of plain 'ol Java objects (POJOs), avoiding the inconsistent, ever-varying, project-specific configuration files.
Spring's power - and the reason why I personally find the very notion of some other pure play IoC container like CDI as a "competitor" to Spring to be laughable at best - is its rich, unified libraries and the platform that it provides through its integrations. Spring provides a clean, consistent metaphor for approaching problems. It provides libraries that are idiomatic and that work wonderfully together, and it provides services that can be weaved in a clean way across an application. These libraries are often layered. They present the 90% case solution first, but provide hooks through the use of things like strategy interfaces and callback methods should you need to tailor the behavior.
The Spring framework is a very good example of the open/closed principle. It is telling that very few people even know about all of the Spring framework's extension points; most people don't need them. For 90% of the use cases, Spring makes the impossible easy, and provides the flexibility to tackle the other 10%. Where possible, Spring makes facets of an application invisible - things that used to require an API. Where there is still an API dependency, Spring typically ships with a cleaner API that insulates you from it or simplifies it.
I mention all of this because Shopzilla's architecture and applications - as much as possible - lived in the 90%; things were as simple as possible thanks to Spring and the incredible diligence of the engineering team. Shopzilla really excelled in all the little 10% spots, too. They used or extended everything, taking nothing for granted. Shopzilla had a custom Spring MVC view resolver and view templating format, for example. They have a custom JMS message listener container that was geared to batching in the style of the Spring framework's message listener container. They made ingenious use of monitoring, Spring's amazing AOP support, JMX, and various open source projects to handle terabytes of logging. The entire federated architecture across several different teams used REST and a standard, Maven controlled archetype to keep projects standardized. In essence, as part of a engineering-wide culture of consistency, and code reuse, Shopzilla had built upon and extended the Spring framework, and it underpinned everything they did. Shopzilla integrated technologies and published reusable solutions for common problems so that everybody benefited in the same way that the Spring framework itself does.
Shopzilla's amazing, and I would recommend it to anybody if you're looking for an amazing company to work for.
So, when I was approached to join SpringSource as the Spring developer advocate, I was at once sad to leave Shopzilla, but elated at the prospect of playing a part in this technology that had so well served me for the better part of the last 8 years. Leaving Shopzilla was painful, of course, but I had to see this opportunity through. I have a deep respect for SpringSource and the committed engineers there who - even as I write this, on the last Saturday on New Year's weekend - are sending emails about developments in the projects! I've been with SpringSource since August. SpringSource is a worldwide organization, so no matter what time of day it is, there is somebody committing something awesome, somewhere, in plain view of the world. If I'm not careful, I forget to leave my desk trying to keep up with all the exciting stuff!
Spring's made my decade much more pleasant than it might've otherwise been. If you'd told me we'd have something like the Spring framework in 2000, and that I'd be privileged enough to work with the people behind it, I would have dropped my CORBA and EJB books on your feet. It would have seemed so cruel to joke about something like that.
It's been a fantastic decade, and a particularly fantastic 2010, and I can't wait to see what comes next.
I hope you had a safe, joyous New Years, and I wish you a fantastic new year, and a new decade.