We can’t stand it, we need to rewrite everything…

This article was originally published in Portuguese in June 2017.

I have heard this phrase several times throughout my career. Software developers know that invariably comes a moment when this kind of discussion comes up, which usually has phrases like: “it’s getting harder and harder to deal with the code”; “if it were to do it from scratch, it would be much faster”; “if we do not rewrite, it will become increasingly slow and dangerous to deal with the code”.

At Locaweb, we had an Email Marketing product for sending and managing email marketing campaigns, which used MongoDB as a database, a nonrelational database known for its ability to store large databases. However, probably because of our limited experience in software development using this type of database and in managing non-relational databases, as the database grew, the system became slower and slower.

So, it was necessary to rewrite the product to use PostgreSQL. We designed this rewrite to be transparent to customers. That is, the client would not be migrated from one database to another, thus avoiding going through a period of unavailability or possible data loss. The rewrite was a success. The entire rewrite process, including putting all existing clients (over 15,000 at the time) into the new database, allowing the MongoDB database to be shut down, took six months, a reasonable time for such an initiative.

However, during these six months, the team delivered nothing new to the customer. No new features. To lessen our customers’ frustration, we decided to hire a third party to build iOS and Android apps on top of existing product APIs. This enabled us to deliver new functionality to our customers while the product team focused on rewriting. If this option did not exist, we would have to find other ways to deliver user value that did not depend on the engineering team.

If you are a software developer, rest assured that if you have not experienced this situation in your career, you surely will. The problem with rewriting software is that by rewriting it, the team stops doing new things for its user, as I showed in the example of Locaweb’s Email Marketing product. From a business standpoint, when software does not evolve, customers see no evolution, may lose interest in using the product and will start looking for better options in the market. Therefore, I would like to make some considerations on the subject.

1) New and better technologies will always appear. In the past, systems were developed with everything in the same code base. Then it was the MVC concept (model, view, controller) separating software code into layers according to their function. More recently, the concept of microservices was created, breaking the application into small, loosely coupled applications, preferably done via APIs and facilitating maintenance. If with every new technology that comes along we are going to rewrite the software, we will certainly do nothing but rewrite software.

2) Software rewriting must have a clear business motivation, ie it must somehow meet the strategic goals of the software owner company as it fulfills a user’s need or solves a customer problem. If there is no clear business motivation, the recommendation is not to rewrite.

3) If rewriting is unavoidable (are you sure?), it is important to think about this rewrite initiative from a business perspective, that is, what is the impact of this rewriting on customers and the software owner. Understanding this is the responsibility of the product manager. Some questions to help you reflect with your team on this impact:

  • What will this rewrite look like?
  • While rewriting is going on, will new features be delivered?
  • How will the new system coexist with the old system?
  • When new features are implemented, will they be implemented on the new system and the old system, or only on the new system?
  • When can new clients be installed on the new system?
  • When will existing customers be migrated to the new system?
  • If there is a proposal to make a synchronizer, so that customer data exists simultaneously on the old system and the new system, what is the cost of this proposal compared to not making this synchronizer?
  • If the difference between the old system and the new system can be perceived by customers, how will this difference be communicated?

There you go, some considerations about software rewriting, a very important topic for any product manager.

Digital Product Management Books

Do you work with digital products? Do you want to know more about how to manage a digital product to increase its chances of success? Check out my new bundle Digital Product Management with my 2 books where I share what I learned during my almost 30 years of experience in creating and managing digital products.

No alt text provided for this image

Leave a Reply

Your email address will not be published.