Weekly Delivery

Apache HBase - a column-oriented datastore

Pros:

+ Fast data load (bulk, progressive or streamed)

+ Fast retrieval of results by row key and/or by column (family, descriptor)

+ Flexible schema, versioned cells

+ Java API and REST API, integration with numerous platforms (Flink, Spark, Hive, Impala)

+ Auto sharding and load balancing in a cluster setup

+ Column-level autorization based on Access Control Lists

Cons:

- Alone it does not offer the features of a RDBMS (e.g. transactions, secondary indexes, query language)

- The need for complementary tools to obtain these features (e.g. Apache Phoenix)

- Denormalized schema

Apache Flink - a platform for distributed analytics

Pros:

+ Broad integration (e.g. HDFS, HBase)

+ High throughput, low latency

+ Support for iterative computations (e.g. machine learning and graph analysis)

+ Supports Java and Scala

Cons:

- Not quite as developed as e.g. Spark

- Does not support Python yet

Hue - a web interface to Hadoop

Pros:

+ Free and Open Source

+ Includes web applications for integration with Hadoop (e.g. HBase, Spark, Hive, Impala)

+ Data exploration

+ SDK

+ Under development

Cons:

- Not a pure visualization tool

- Limitations when it comes to visualization and data handling

- No support for multimedia

By Oscar Ivarsson

More from Oscar Ivarsson