Teaching a fish to fly

Deciding how to migrate a software system depends on what you are starting from. So if it seems like you want to teach a fish to fly, you need to stop and ask yourself if that's really what you want.

A common idea is that applying all the current buzzwords to your system will make it better. Add CI\CD, Kubernetes, DevOps, microservices, etc, etc. But what does all that get you, and at what cost?

It all depends on identifying your bottlenecks. If you are developing so many features that you can't push them to production fast enough, then CI\CD makes sense. Otherwise, focus on where your bottleneck is. Testing? Agile process? Scheduling? Work on those before starting a major overhaul. The great thing about software is that you can make a fish fly, but you still need to answer why.

Software is not a Field of Dreams, just because you built it, they may or may not come. Build the system you need first, and identify areas for improvement. Then build to address those areas. But if your system is operating fine as is, has enough capacity, and you have a plan to update and upgrade, don't mess with it.

Comments

Post a Comment

Popular posts from this blog

The Power of Log

I recently started a new position at a new organization, and it was the biggest change in my career. My family moved overseas to Cambridge, leaving behind all our friends and most of our belongings. But working on a fixed-length contract is a new opportunity for me. It let me answer the old question "Where do you want to be in three years?" knowing that in three years I'll need to find a new job, instead of just thinking about it. I've been a tech lead for many years, mostly because being a development manager didn't really appeal to me. So I started looking at other positions I would be qualified for, if I had the relevant experience. I decided I've got three years to get that experience, and plenty of opportunities as my new organization. But looking at all the things I will need to do is a daunting list, and I needed a way to break it down into SMART-type goals. The first thing I noticed is that 3 years is about 1,000 days. So that gave me a deadline fo...
Read more

Welcome to the Pessimistic developer

Hello and welcome to the Pessimistic developer. I want to take a closer look at so called best practices and see if they are just common practices. Or to put it another way,  i want to use a critical eye and see if common patterns are really the best choice for you and your project. So I don't want to be the guy giving a TED talk, saying how great the newest technology is. I want to be the guy giving the talk recommending caution and making sure it's the right fit for you. Modern stacks give us the ability to fit just about any square peg into any round hole, but now it's our job to decide if that's the right fit. I think the core of this problem is that developers have not paid attention to the principle of " You're not going to need it " and have added too much complexity to systems that could really be implemented more simply. But architecture is just as much a part of the design as the code itself, and it should be implemented as simply as possible u...
Read more

We need to talk about Java

Has there ever been a bigger elephant in the room than Java? Bigger in terms of marketshare, and bigger in terms of  footprint size. I started out as a Java developer, and have done most of my work with it, but I definitely feel that there are better tools for specific jobs. If you want one single language to do it all, maybe with some HTML thrown in, Java will do it. But how many times is that really the case? Java primarily suffers from what I call Unintended Dependencies, or Why Do I Need Your X to Build My Y? As everyone knows, Java ships with all the libraries to do everything from application UIs to database persistence to messaging to system calls, to building native platform calls, and more. I've been around long enough to remember when you needed to add J2EE as an additional package, and I think that's when things started to go wrong. Now, just to build a Docker image of OpenJava it takes about 500 megabytes of disk. Add Tomcat onto that plus whatever app you're d...
Read more