index.md

layout	sidebar	title
page	hidden	Kite: A Data API for Hadoop

Kite is a high-level data layer for Hadoop. It is an API and a set of tools that speed up development. You configure how Kite stores your data in Hadoop, instead of building and maintaining that infrastructure yourself.

High-level tools

Kite's API and tools are built around datasets. Datasets are uniquely identified URIs, like dataset:hive:ratings.

Dataset is a consistent interface for working with your data. You have control of implementation details, such as whether to use Avro or Parquet format, HDFS or HBase storage, and snappy compression or another. You only have to tell Kite what to do; Kite handles the implementation for you.

{% highlight text %} $> kite-dataset csv-import ratings.csv \ dataset:hbase:zk/ratings Added 1000000 records to dataset "ratings" {% endhighlight %}

Kite's command-line interface helps you manage datasets with pre-built tasks like creating datasets, migrating schemas, and loading data. It also helps you configure Kite and other Hadoop projects.

  Install the Kite command-line tools or start Kite's CSV tutorial

{% highlight java %} View latest = Datasets.load(uri) .from("time", startOfToday) .to("time", now); {% endhighlight %}

Kite's data API provides programmatic access to datasets. Using the API, you can build applications that directly interact with your datasets. For example, you can load a dataset and select a subset of it for a MapReduce pipeline.

  Learn more about the Kite API

Low-level control

When you create a dataset, you control your data layout, record schema, and other options with straight-forward configuration. Then you can focus on building your application, while Kite handles data storage for you. Kite automatically partitions records when writing and prunes partitions when reading. It will even keep Hive up-to-date with a dataset's newest partitions.

{% highlight text %} time,rating,user_id,item_id 1412361369702,4,34,18865 ... {% endhighlight %}

{% highlight json %} [ {"type": "year", "source": "time"}, {"type": "month", "source": "time"}, {"type": "day", "source": "time"}, ] {% endhighlight %}

{% highlight text %} datasets/ └── ratings/ └── year=2014/ ├── month=09/ │ ├── day=01/ │ ├── ... │ └── day=30/ ├── month=10/ │ ├── day=01/ │ ├── ... {% endhighlight %}

  Learn more about Kite datasets

Configuration-based transformation

Kite morphlines is a flexible way to express data transformations as configuration.

  Go to the Morphlines introduction or reference guide