The File Storage Challenges for Quantitative Trading Teams
- More and more quantitative trading teams are building their own trading, analysis and backtesting systems in public cloud or hybrid cloud. The storage products and services need to adapt to the cloud environments;
- Quantitative analysts might use various programming languages and data processing tools, such as Python, R, Java, MATLAB, SAS, etc.; Data types involved are diverse, including CSV, Parquet, TXT, Excel, HDF, etc.; Thus the storage platform needs to have the ability to store various types of data.
- Quantitative research will use a large number of machine learning and deep learning tasks, which pose new challenges to the throughput requirements of storage systems.
- Kubernetes is becoming the de facto standard for resource orchestration. The file storage also needs to be friendly to the Kubernetes ecosystem.
- Control of modular access and intellectual property protection, as well as data security issues are essential requirements for quant research.
- JuiceFS is designed for cloud environments, suitable for deployment in public, private and hybrid cloud architectures.
- 100% compatible with POSIX, HDFS, S3 APIs. You can build applications using your familiar tools and frameworks and migrate legacy application code without minimal cost.
- The throughput of JuiceFS can be linearly increased with the number of clients, providing high throughput access for hot data in quantitative research.
- JuiceFS can be used in Kubernetes through CSI Driver and is compatible with various Kubernetes distributions and cloud hosting services.
- JuiceFS supports data encryption, which can guarantee analysts’ research privacy.