Note In this article, I elaborate on the risk of vendor locking and how careful solution design helps prepare for the emergency. To be honest, that is not the main benefit of owning a business-tailored IoT solution, and this topic emerged on its own during the writing process. First, I considered it a side track, but then I realized it is an important perspective to share with you. Enjoy this text and share your experiences with me!

The Prologue In the beginning, no one really knows what they truly need. We call it the discovery phase when business domain knowledge clashes with the solution design mindset. We discuss the business aspects and random technical capabilities that may or may not be relevant to a given case. You start questioning the reason for this initiative, and you might be right. At the end of a few cycles, there are no tangible outcomes.

Recently, I had to significantly redesign the database structure of an IoT platform I designed one month ago. Changing the database forced the adaptation of the backend logic*, which demanded my extra attention and careful testing. During that change, I was constantly questioning my initial reasoning and my ability to design solutions. “That was obviously a wrong decision! Why does that setup not support this access pattern?!?!" Once my frustration level dropped a bit, I had yet another reflection - “There was no way I could have predicted all of that one month ago”.

The initial connectivity of IoT devices to the AWS Cloud can be a major hurdle, impacting user experience and project success. Every device is unique, and ensuring secure, seamless onboarding for thousands of units presents significant manufacturing and operational challenges. I’ve seen my customers struggle with these issues for years. Finally, I decided to encode my experience and develop a device provisioning module that simplifies this process. It allows you to securely connect any device to your AWS account without requiring expertise in AWS or a large IT department.

There is no technical debt, only an investment opportunity in education or future business returns. The effort of fixing the consequences of bad decisions is part of the price you have to pay. The only loss happens when you fail to acknowledge the “debt” and choose to pretend it is not there. I designed and built numerous IoT systems, and I made countless mistakes in the process. The truth is that when you start creating something innovative, you have no clue if your design decisions are correct or not; there is no way to predict the long-term impact of the way you structure the early version of your system.

Amazon Timestream for LiveAnalytics will no longer be open to new customers starting June 20, 2025. That is very bad news for numerous AWS IoT deployments. I deployed Amazon Timestream for various customers using this time series database to store telemetry data. I liked Timestream because it was an AWS-managed service, so there was no need to worry about the scalability and availability of the underlying infrastructure. Unfortunately, AWS decided to disable this service for new customers.