This session will describe the advantages and architectural considerations of using Cloud Functions, IBM’s serverless offering, when designing, implementing, and deploying a scalable data science & analytics project.
When you hear the term serverless, what comes to mind?
Eponymously, functions-as-a-service implies the invocation of relatively compact and short-lived pieces of code, and not computationally intense data science workloads.
Anyways, we have various flavors of Apache Spark and Kubernetes for heavy-duty work, right?
However, while powerful, these frameworks impose complex cluster management and configuration, as well as financial costs, on DevOps teams who simply want to Get Stuff Done!
This session describes the use of Cloud Functions, rather than Spark or K8s, to handle a scalable energy modelling & forecasting workload as part of a pan European project underway at IBM Research Dublin entitled GOFLEX.
The benefits of this approach, including automated scalability, backend simplicity, and cost optimization are discussed, as well as architectural and development considerations.
YOU MAY ALSO LIKE:
Pushing Data Science & Analytics onto IBM Cloud Functions
Robert is a software engineer based in the IBM Research Dublin Lab, and applies the Taoist principle of wei wu wei to software development whenever possible.