Vague. Unchallenged. Hard to argue against. FUD.
Example 1: Lets build a CMS for our forms for “Extensibility”. Business people can add new fields easily.
Result: Business people never used it. When they had to, they would have a developer sit right beside them and do it. Maybe all we needed was a simple Developer guide to add a new field in few hours, instead of a point-and-click interface?
Example 2: Lets design a big Database layer for easy “Configurability”. We should be able to switch database in a single Magic file.
Result: In 10 years, I’ve seen only one business make serious effort to completely swap a fully vested database. And when it happened, the “Magic file” did not help. There was so much operational work. Incompatibilities and gaps in functionality. And the client asked us to switch “one half” of our models to the new NoSQL database. We tore our hair apart — our Magic toggle was a single point of change, but this was cross-cutting.
In today’s world, we’re well past a point where there is no way to design a single configurable layer for modern document/KV stores (e.g. Redis/ CouchDB/ DynamoDB/ etc). Not even SQL Databases like Postgres/ HSQLDB/ SQLite are compatible for that matter. Either you completely dumb down your data layer (and struggle with delivering functionality), or acknowledge the database as part of your solution (e.g. postgres geo/json features) and throw away configurability guilt. Your stack is as much part of your solution as your code. When you let go of this vague X-ity, better solutions start to emerge. e.g. You can now break data access vertically (small DAOs for each action) instead of horizontally (magic configurable layer), or even pick and choose different data stores for different functionality [micro] services style.
Example 3: We built an OAuth system for enterprise clients. For the Internal administrators — we were asked to use a secondary Google OAuth system. Because Security. If someone hacks our OAuth, business didn’t want them to get access to admin credentials. Google OAuth is more secure, and who can argue against “more security” at any point?
Result: If someone really wanted to hack into our system, they don’t have to go through the OAuth layer. We had many vulnerabilities lying around. e.g. they could have just done privilege escalation. So all that effort of supporting two different OAuth user profiles and systems everywhere had little to no returns in securing our system, compared to properly securing our base first.
TL;DR — Don’t let <X>-ities go unchallenged. Clearly define and evaluate the Scenario/Story/Need/Usage.
Tip: Ask a simple question — “What’s an example story/scenario?” — and then dig deep on that scenario. This exposes flaws in most <X>-ities.
via Medium http://ift.tt/2ahQ2IO