Accesto

Time and budget constraints during the MVP phase often mean compromises in quality of the development. Testing is done on small datasets and few users. As the application gains traction and traffic increases - performance drops to disappointing levels. According to Google just 0.5 second of additional load time may mean a drop of traffic by 20%. Amazons research shows that every 100ms of latency cost them 1% in sales. The immediate response to performance is often buying more server power, but this is costly and short-term. Our approach starts with analyzing the code performance and identifying bottlenecks. By using specialized tools, we can do it on production servers. Usually within just days we can offer quick fixes that significantly improve the performance. In the long run we optimize the code, architecture as well as use Docker and Kubernetes to optimize server performance with lower bills from the cloud service. In the post-MVP phase businesses often wants to scale the app and go global, but the technology behind the product is not ready. Problems include one server with users around the world, multiple features for few users etc. The solution is to reject costly vertical scalability (increasing the number and size of servers) and promote horizontal scalability (dynamic management of cloud capacity). With most businesses experiencing peak traffic and relatively calm periods, sensible management of resources is crucial for profitability of the operation. Old code, technical debt and outdated stack is difficult to expand, and extra effort involved becomes an interest the business has to pay. Technology locking is another problem - by sticking to old versions of programming languages, not using modern frameworks and even using modern technology, but rare and expensive, the availability of talent is restricted which impacts the cost of employment. Sometimes software development companies advise radical measures such as scrapping the old application and building a completely new one in parallel. But this is ...