Why this post?
Partly, this blogpost is a result of a promise to Lukas Eder. Basically my vision adheres quite nicely to the “Thick Database” driven by Bryn Llewellyn and Toon Koppelaars who, understandibly, drive this from an Oracle perspective.
It –more than of course- also nicely fits EnterpriseDB or even vanilla PostgreSQL database landscapes.
There is apparently still so much confusion in the world on the how, why and what of good application development and architecture that I decided to chip in my bit. I think I have a bit of an idea on how this aught to work and I also think it is not a half bad idea, plus a couple of people whom I highly regard, seem to agree with it. So here goes…
Traditionally there is no big love between application developers and their persistence-store. I don’t really know why because I never found the opportunity to do a real inquiry, but I think I have a reasonable understanding.
Basically there is constantly the enormous pressure of delivering new features and functionality. So much even that the basic development work, the more “boring” and “time consuming” things -why pay now, if you can also pay later- get postponed. Things like peer-reviews, (integration) testing, technical design… Basically, more people means more features.
If even these things get too little attention, why would something like an overpriced library-box get more attention? Not to mention these DBA’s you need to pass to even get close(r) to this library-box…
Here are my four reasons why I think
it should deserve a chance!
Plain and simple. It is easier. If you take a structured query language like PL/(pg)SQL, it is basically easy. Founded on the programming language ADA, it is easy to understand and one can quite easily build a number of routines, procedures, packages, functions to let the database chop and glue data and just deliver the results for your application to consume.
It saves you the time of having to (re)write some of this more complex stuff in your application or perhaps even over a several applications.
As said under the first point: build once, use many times. By creating mechanisms in the database, you get the opportunity to think about the separating data manipulation mechanisms from data representation mechanisms and where you want to put which specific function. Of course, this decision process takes a little extra time in the beginning, but will repay many times over as your projects grows and gains meaningful complexity (is there something like “meaningful complexity”, well, yes…).
Quicker also is in operational response times. Querying a stored procedure will bring agility to your application in a sense that this stored procedure will be much quicker in getting you the answer than if you do this in a distributed (middle-ware) environment. These stored procedures can be accessed through REST-endpoints, giving flexibility and the possibile desired disconnect between the database and the application layer.
Powerful, because you are and you remain close to the data. No data transportation overhead, no latency. These kinds of slimming down, mean less requirements to infrastructure and distributed capacities (either hardware or (virtual) “cloud infrastructure”. This slimming down frees up budget which can then be spend on the more meaningful bits and pieces of your application.
4. More consistent
Finally, I think this approach brings more consistency. As you do things near the data it lives on, near the processing power you depend on, you also get single access paths to specific bits of your data, to specific constructs that drive the value of the application you are building. Through this, it does not matter from where you call this service, REST-endpoint, stored procedure, or whatever you call it, you always get the same answer, driving a consistent decision process based on that application.
And if ever something changes, there is also just one place for you to ensure these new requirements are added. Voila, consistency throughout your application landscape, as all that depends on this data-set gets this uniquely updated information.
The magic of working together
There is a lot of misunderstanding between DBA’s and Developers. I have been in both roles at some point and I have seen this happen first hand. One of the things, though, in that force field, I have learned, is the power, the joy and, through this, the magic of working together.
In the end, we all have the same goal, which is furthering our business by being the best at what we do. This means, for a developer, meeting feature request, short development cycles, quick delivery and as much as possible, get it right in one go. For a DBA this means making sure the database stays consistent, performant and available. And, in extension of that, for operations it means that the final product must be easily and quickly deployable to enter into an uneventful and dependable life-cycle.
Bringing together these seemingly conflicting disciplines is fun! By investing a little time in exploring the other disciplines, you will find common drivers, in a sense that everyone want the same thing. By getting over nearly religious initial differences, you will find magic in the combination. You will reach your goals earlier with the bonus that your co-workers will also reach their goals earlier and have a better end result than you dreamt possible.
Coming from another world, I do not know the requirements for becoming a JAVA-Champion. I imagine it to be not too much different from other recognition programs out there… But…
If you create more features and functions and you are able to run your application with greater concurrency on (way) less platform-power. Thus increasing the RIO on your application, this makes you a true Champion of JAVA and your business!! If you are able to combine this with some magic in your cooperation’s…
Believe me, it is more fun in the end too!