My understanding is that microservices is splitting the code into modules but can be very complex themselves and lambda functions are much more unit level, and their uniqueness is in how they're deployed
One might consider an application stack that was previously a monolith and hence difficult to update or change without a large 'blast radius' or impact to all components; i.e. downtime during a patch or upgrade. Microservices allow for de-coupling and changes made to components, and far greater introspection (since those services are labeled and give off unaggregated metrics / traces / logs).
The serverless architecture takes away the burden of maintaining the server at the expense of degraded latency on the initial invocation of the service. Why? The cloud provider provisions the server on-demand when the service is invoked unless there were previous invocations of the same service.
It's also worth noting that serverless architecture is usually billed by the time of code execution. So, computationally expensive operations on serverless architecture might not be the best choice.
Good work as usual Justin. Thanks man!
Thanks - this was very helpful for a non-technie like me.
Is serverless related to microservices and how?
My understanding is that microservices is splitting the code into modules but can be very complex themselves and lambda functions are much more unit level, and their uniqueness is in how they're deployed
One might consider an application stack that was previously a monolith and hence difficult to update or change without a large 'blast radius' or impact to all components; i.e. downtime during a patch or upgrade. Microservices allow for de-coupling and changes made to components, and far greater introspection (since those services are labeled and give off unaggregated metrics / traces / logs).
Great post!
The serverless architecture takes away the burden of maintaining the server at the expense of degraded latency on the initial invocation of the service. Why? The cloud provider provisions the server on-demand when the service is invoked unless there were previous invocations of the same service.
It's also worth noting that serverless architecture is usually billed by the time of code execution. So, computationally expensive operations on serverless architecture might not be the best choice.