As the use of Amazon Web Services (AWS) has become mainstream, one service that gets a lot of attention is Amazon’s Redshift database. After all, Amazon Redshift is a fast, fully managed data warehouse that makes it simple and cost-effective to analyze data using standard SQL with existing Business Intelligence (BI) tools. Users can run complex analytic queries against substantial amounts of structured data, using sophisticated query optimization, columnar storage on high-performance disks, and massively parallel query execution. Plenty of well know companies are already using AWS Redshift ranging from Liberty Mutual Insurance and Pinterest to Yelp and Nasdaq and many more.
However, despite its obvious enterprise benefits, there are a couple of quirks with Redshift that are cropping up on forums and noticeboards. Amongst these, some main themes have emerged:
- Query performance could be better: In a world where time is money, Redshift is often not able to deliver the absolute query runtimes that many users would love to see.
- System resources are easily overwhelmed: Only a handful of concurrent requests is all that is needed to saturate resources, resulting in severe performance degradation.
- No access to underlying OS: Because of the fully managed licensing model, there is no underlying access to the OS and storage.
- Sizing and pricing is often overkill: There are only two types of instances for SSD or HDD storage and entry-level costs are relatively high.
How Brytlyt can help companies to get more from Redshift:
Brytlyt is based on PostgreSQL and uses Graphics Processor Units (GPU) to massively accelerate SQL queries. Brytlyt compliments a Redshift deployment and can fill some of the gaps that exist with standalone Redshift. The most important thing to remember is it is a PostgreSQL fork, just like Redshift, so the two systems are virtually identical from a user perspective.
- Brytlyt solves the biggest problem facing Redshift users today – Performance. Expect a 20x to 100x improvement, especially in scenarios where many users run queries at the same time. A single line of code is all that’s needed to connect Brytlyt to Redshift, and with Brytlyt’s basis in PostgreSQL, any code originally developed for Redshift will run out-of-the-box on Brytlyt.
- The Brytlyt licensing model gives the user access to the OS and configuration so they can tune the system to suit their specific workloads.
- With Brytlyt, users can add and remove GPU resources at will, scaling their processing capability to suit the company’s needs. There is flexibility for entry-level workloads that require only one GPU, all the way up to enterprise workloads that require hundreds of GPUs.
- The Brytlyt platform also comes with the SpotLyt interactive workbench for real-time visualization of billion row datasets and interactive data exploration. So not only can businesses continue to use existing analytics, SQL code, and data visualizations, they can also use SpotLyt to interact with their data.
It is easy to use Brytlyt to accelerate Redshift workloads. The data below is from independent benchmarking done by Mark Litwintschik. http://tech.marksblogg.com/benchmarks.html
Brytlyt is a PostgreSQL clone, just like Redshift, so it fits neatly into any Redshift deployment.
References:
