Skip to content

Tech

How Alphaus saves on costs by "stitching storage"

(Repost from https://flowerinthenight.com/blog/2024-07-24-spillover-store/)


One of Alphaus' data processing pipelines ingests around 10TB of client financial data per day. The processing engine is running on GKE with around 80-100 (depending on what week of the month) pods sharing the total workload. Each pod has around 10GB of memory and 30GB of attached storage. The consistency of this load allowed us to purchase enough Committed Use Discounts (CUDs) for the underlying VMs to save on compute costs.

bluectl

In today's post, I will introduce bluectl, the official command line interface (CLI) for Alphaus services. To those who prefer CLIs (that includes me), we provide bluectl for interacting with our services. It uses the same API that powers our Ripple/Wave[Pro]/Aqua UI consoles.

Blue API

In today's post, I will talk a little bit about Blue API, our public-facing, as well as internal APIs. Blue API allows you to programmatically access our services such as Ripple and WavePro. It uses protocol buffers for its service and message definitions, and gRPC for implementation and server/client stub generation. It also uses grpc-gateway for proxying JSON/REST requests to gRPC, and generating OpenAPI documentation. This way, you have the option to use our APIs using either gRPC or JSON/REST, or both.