Hello, Kite SDK 1.0

Categories: Kite SDK

The Kite project recently released a stable 1.0!

This milestone means that Kite’s data API and command-line tools is ready for long-term use.

The 1.0 data modules and API are no longer rapidly changing. From 1.0 on, Kite will be strict about breaking compatibility and will use semantic versioning to signal what compatibility guarantees you can expect from a given release. For example, breaking changes require increasing the major version number, so both minor and patch updates are safe to use without code changes, and binary compatible.

Kite provides some additional guarantees as well:

  • Kite’s command-line tool, kite-dataset, also follows semantic versioning. Changes that may break scripts will require updating the major version number.
  • Incompatible changes to Kite’s on-disk formats, Avro and Parquet, will also be signalled by a major version update.
  • Kite’s storage formats, metadata, and file layout are forward-compatible for at least one major version. You can update Kite, write to existing datasets, and roll back to the previous version safely.

We’re excited to get the Kite 1.0 release out. Now, Kite provides a great high-level API built around how you work with your data, and stability guarantees so you can be confident building on top of it.

Learn more about Kite:


2 responses on “Hello, Kite SDK 1.0

  1. Jason

    Hi there, question about Kite Datasets! We’ve written an HBase DAO utilizing Kite Datasets and alas, they want to turn on Kerberos for the cluster (CDH5.5). My question is, which user does Kite attempt to read/write as? I always thought it would be the current system user, but looking at the HBase data in Hue, I see everything is created by “hbase” which leads me to believe Kite might be defaulting the user as “hbase”. I couldn’t find evidence of this in the source or documentation. Perhaps you might be able to point me in the right direction or confirm. Thanks in advance.

  2. Aniruddh

    We have a project where we will implement a Data Lake and doing transformations using Hadoop Stack (like Hive and Spark). We don’t have an explicit Java Client so far who will be user of this data. Kite SDK has another interesting feature for schema evolution.
    Can you please suggest if you have seen anyone using Kite SDK in Data Lake. If yes then exactly for what use case