A year ago, we migrated ReductStore, time series database for unstructured data, from C++ to Rust. Now, reflecting on that decision after a year, it's clear that it was the right move, yielding the following benefits:

• Fewer bugs related to memory management, move semantics, and threading
• Easier codebase porting to MacOS and Windows
• Better dependency management with Cargo compared to what we had with C++

Today, I'm pleased to announce that we've rewritten the CLI client from Python to Rust. Our primary motivation was distribution. Although Python is one of the most widely used programming languages, we don't want to require users to install an interpreter to use our tools, and standalone installers are too bulky for us. With Rust, we can build compact and blazingly fast executable binaries for most popular platforms.

wget https://github.com/reductstore/reduct-cli/releases/latest/download/reduct-cli.linux-amd64.tar.gz
tar -xvf reduct-cli.linux-amd64.tar.gz
chmod +x reduct-cli
sudo mv reduct-cli /usr/local/bin

In edge computing, managing time series blob data efficiently is critical for performance-sensitive applications. This blog post will compare ReductStore, a specialized time series database for unstructured data, and MongoDB, a widely-used NoSQL database.

Using Docker containers for straightforward setup, we'll examine the speed of each system. We'll go through setting up ReductStore buckets and preparing MongoDB collections, focusing on how to effectively store and access blob data for time series scenarios.

By conducting performance tests on binary data insertion and retrieval, we aim to provide insights into which system might best serve your application's needs.

For those interested in replicating our benchmarks or conducting their own evaluations, we've made our methods easily accessible through this repository.

TimescaleDB is an open-source time-series database optimized for fast ingest and complex queries. It is engineered up from PostgreSQL and offers the power, reliability, and ease-of-use of a relational database, combined with the scalability typically seen in NoSQL systems. It is particularly suited for storing and analyzing things that happen over time, such as metrics, events, and real-time analytics.

Since TimescaleDB is based on PostgreSQL, it supports blob data and can be used to store a history of unstructured data such as images, binary sensor data, or large text documents. In this article, we will use the database as a time-series blob storage and compare its performance with ReductStore, which is designed specifically for this use case.

TimescaleDB and ReductStore both have Python Client SDKs. We'll create simple Python functions to read and write data, then compare performance with different blob sizes. To repeat these benchmarks on your own machine, use this repository.