Faas Best Practices:
Posted: Sat Apr 19, 2025 6:05 am
1- Reuse connections (HTTP, database, etc.) that were established in previous function calls. 2- Use environment variables to pass parameters to the function. For example, to configure access to storage. 3- Have control over your function's libraries, avoid belarus mobile database and updates that change some behavior. 4- Reduce the complexity of dependencies, prefer simple and fast-loading frameworks. 5- Avoid recursive code (the function calling itself), because of the risk of loss of control in function calls.
Some critical points of FaaS:
Below, we highlight the critical points of using FaaS, so that it is clear whether its use is strategic for the business, even with some obstacles: 1- Being tied to a single cloud provider, since the companies we mentioned provide the FaaS implementation as they see fit and changing clouds can be complicated. 2- “Cold Start” phenomenon in which a function takes a long time to start responding. This occurs in two situations. When you have just published or updated a function, because the cloud is bringing up the execution environment. In the second case, your function has not had any calls for a long time and the execution environment ended up being closed by the cloud provider. 3- You pay per call - the good side of the low price in Production, can end up being a high price during Development, Testing and Homologation, because you are paying for each call in the same way. 4- Each provider chooses how the logs will be collected or displayed and may have a hidden cost for using monitoring services. FaaS is promising, as it diversifies the options for improving application development. In a future article, we will discuss another view of the FaaS model linked to Docker. Follow our blog and stay up to date with the latest news.
Some critical points of FaaS:
Below, we highlight the critical points of using FaaS, so that it is clear whether its use is strategic for the business, even with some obstacles: 1- Being tied to a single cloud provider, since the companies we mentioned provide the FaaS implementation as they see fit and changing clouds can be complicated. 2- “Cold Start” phenomenon in which a function takes a long time to start responding. This occurs in two situations. When you have just published or updated a function, because the cloud is bringing up the execution environment. In the second case, your function has not had any calls for a long time and the execution environment ended up being closed by the cloud provider. 3- You pay per call - the good side of the low price in Production, can end up being a high price during Development, Testing and Homologation, because you are paying for each call in the same way. 4- Each provider chooses how the logs will be collected or displayed and may have a hidden cost for using monitoring services. FaaS is promising, as it diversifies the options for improving application development. In a future article, we will discuss another view of the FaaS model linked to Docker. Follow our blog and stay up to date with the latest news.