AI as a Service: Serverless machine learning with AWS by Peter Elger & Eóin Shanaghy

Author:Peter Elger & Eóin Shanaghy [Elger, Peter & Shanaghy, Eóin]
Language: eng
Format: epub
Publisher: Manning Publications Co.
Published: 0101-01-01T00:00:00+00:00

Figure 6.6 The canonical serverless CI/CD architecture is part of SLIC Starter. It uses a CodePipeline pipeline for each module. The execution of these pipelines in parallel is coordinated by an orchestrator pipeline. The build, deployment, and test phases are implemented as CodeBuild projects.

How much we can understand them depends on the interfaces those services provide to report their status. The degree to which systems report their status is called observability. This term is increasingly being used instead of the traditional term, monitoring.

Monitoring versus observability

Monitoring typically refers to the use of tools to inspect known metrics of a system. Monitoring should allow you to detect when problems happen and to infer some knowledge of the system. If a system does not emit the right outputs, the effect of monitoring is limited.

Observability,2 a term from control theory, is the property of a system that allows you to understand what’s going on inside by looking at its outputs. The goal of observability is to be able to understand any given problem by inspecting its outputs. For example, if we have to change a system and redeploy it in order to understand what’s going on, the system is lacking in observability.

One way to think about the difference between these two terms is that monitoring allows you to detect when known problems occur, and observability aims to provide understanding when unknown problems occur.

As an example, let us suppose that your application has a well-tested, working sign-up feature. One day, users complain that they are unable to complete sign-ups. By looking at a visual map of the system, you determine that errors in the sign-up module result from failures in sending sign-up confirmation emails. By looking further into the errors in the email service, you notice that an email sending limit has been reached, preventing the emails from being sent. The visual map of dependencies between modules and errors led you to the email service logs, which gives the root cause details. These observability features helped to resolve an unexpected problem.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.