Tessellate is a command line tool for reading and writing data to/from multiple locations and across multiple formats.
This project is under active development and many features are considered alpha.
A primary activity of any data-engineering effort is to format and organize data for different access patterns.
For example, logs frequently arrive as lines of text, but are often best consumed as structured data. And different stakeholders may have different needs of the log data, so it must be organized (partitioned) in different ways that support those needs.
Tessellate was designed to support data engineers and data scientists in their efforts to automate managing data for use by different platforms and tools.
Tessellate is a command line interface (cli).
It can read or write files locally, or from HDFS or AWS S3.
It also supports most text formats and Apache Parquet natively.
And during writing, it will efficiently partition data into different paths based on input or derived data availble in the pipeline.
Tessellate may be used from the command line, or in a container.
It also natively supports the Clusterless workload model.
Other uses include:
Data inspection from a terminal
Host log processing (push rotated logs to HDFS or the cloud)
For cloud processing arriving data (like AWS Fargate or ECS via AWS Batch)
As a serverless function (like AWS Lambda) [we plan to publish artifacts in Maven for inclusion in Lambda functions]
Tessellate uses Cascading under the hood for all of its processing.
Historically, Cascading has been used to run large and complex Apache Hadoop and Tez applications, but it also supports a local execution without Hadoop runtime dependencies. This makes it very suitable for local processing, or in a cloud environment on AWS ECS/Fargate or AWS Lambda.