December 2015 Summaries

The latest version 3.0.0 of the DataStax C# Driver has been released, supporting Apache Cassandra 3.0. This release includes support for new schema metadata introduced in Cassandra 3.0 and maintains compatibility with earlier versions. It also introduces Materialized Views metadata information and a new Index metadata API. Performance improvements under .NET runtime have been implemented, resulting in significantly higher throughput rates compared to v2.7 of the driver. The default settings have been revamped to recommended values. Future versions will use pure semantic versioning while maintaining compatibility with earlier Cassandra versions. Feedback is welcome and can be provided using the following link: Version 3.0.0 of the DataStax C# Driver is now available on Nuget.

Cassandra is a popular choice for time series data storage, which includes messages, events or similar transactions with a time element. It offers various techniques and tips to improve application performance both now and in the future. Choosing long term storage requires understanding business requirements, as some applications may need to hold data for longer periods such as seven years in financial companies. Time series examples include credit card accounts, smart meter energy data, and tick data for financial instruments. Cassandra's clustering columns can be used to store time series data efficiently, but it is essential to consider storage capabilities and implications of having billions of columns in tables. Modifying the data model structure as its usefulness decreases is recommended, and more information on Cassandra and data modeling can be found at DataStax Academy's tutorials and certification options.

Apache Cassandra 3.0 introduced a significant refactor of its storage engine, with changes made to better align the internal structure of objects manipulated by the storage structure with what is exposed through CQL (Cassandra Query Language). The new Map<byte[], SortedMap<Clustering, Row>> structure replaces the old Map<byte[], SortedMap<byte[], Cell>>. This change allows for more efficient handling of deletions and a more compact storage format. Additionally, the use of COMPACT STORAGE is discouraged in C* 3.0 as it no longer offers any advantages over non-compact tables.

DataStax DevCenter 1.5 is now available, offering compatibility with Apache Cassandra® 3.0. This version introduces support for Materialized Views and Multiple Indexes, along with content assist, quick fix suggestions, validations, and wizards. The update also includes numerous improvements and bug fixes. Key features include automatic primary key specification when creating Materialized Views using the Wizard, ALTER wizard for existing views, and enhanced support for multiple indices on the same column. Additionally, DevCenter 1.5 simplifies dropping User-defined Types, Functions, and other schema elements by providing context menus with drop options. The update also improves date formatting to match cqlsh and introduces new icons for schema entities.

Cassandra repairs involve comparing data between replica nodes, identifying inconsistencies, and streaming the latest values for mismatched data. Repairs are resource-intensive, requiring CPU to generate Merkle trees and networking/IO to stream missing data. OpsCenter's repair service splits up jobs into smaller tasks and runs them continuously, reducing manual workload and spikes in resource usage. Repair sessions are identified by a UUID, and logs from healthy repairs show only INFO messages with no WARN or ERROR. Repairs can fail due to networking issues or sstable corruptions, which may require admin intervention such as running nodetool scrub. Regularly scheduled repairs help maintain cluster health and prevent data inconsistencies.