March 2016 Summaries

DataStax Enterprise (DSE) offers search indexing capabilities for Apache Cassandra™ data, eliminating the need for double writes or ETL's between separate DBMS and Search clusters. The API call enables full text/fuzzy search, adhoc Lucene secondary index powered filtering, and geospatial search. Advanced tuning options are available to optimize search indexing latency and search query latency. Real-Time (RT) indexing in DSE Search provides faster indexing performance with 4.7.3, especially for the tails of the latency distribution. Monitoring tools like OpsCenter and JMX provide insights into search performance metrics, while tips such as using filter cache, partition routing, and enabling docvalues can improve query performance. DSE Search also offers unique features like query retries on node failure and distributed query routing decisions based on node health.

The C/C++ driver for Apache Cassandra has released version 2.3, which includes support for all features of Apache Cassandra 3.0, including materialized view metadata. This release also introduces new load balancing policies and full support for Apache Cassandra 3.0. Additionally, the driver now supports inspecting metadata for materialized views, allowing retrieval from either a keyspace or its base table. The materialized view metadata is almost identical to table metadata, with the exception of the ability to retrieve the base table using cass_materialized_view_meta_base_table(). This release also includes several community-contributed features and encourages continued feedback and involvement from users.

CQL and DSE Search offer scalable, highly available, operationally simple, and user-friendly access to a Lucene-backed index. Developments in DSE 4.8 point releases have made it possible to query search indexes directly from CQL using DataStax drivers, eliminating the need for two sets of DAO's, load balancers, and addressing performance issues related to HTTP Basic Authentication. NumFound and facet queries are now supported via CQL and solr_query in DSE 4.8.1+ versions. Additionally, partition routing is a useful multi-tennant feature that limits the amount of fan out for search queries, while JSON queries enable advanced searching through CQL. The timeAllowed parameter allows users to set maximum latency for search results when scanning entire datasets is not required.

Cassandra 3.0.4 and 3.4 introduces sstabledump, a new utility for exploring SSTables. It replaces sstable2json, which was removed from Cassandra in version 3.0. The new tool can export SSTable content to human-readable JSON format. In contrast to previous versions of Cassandra, SSTables are now composed of partitions and their rows, with metadata such as clustering key values, timestamps, and TTLs defined at the row level. This change leads to a more understandable representation and a reduced storage footprint for SSTables. sstabledump supports Cassandra 3.X SSTables and can be executed outside of the Cassandra environment without requiring cassandra.yaml in the classpath. The utility provides both JSON and 'debug' output formats, with the latter being more concise and convenient for advanced users to analyze SSTable contents.