October 2015 Summaries

The C/C++ driver for Apache Cassandra has released version 2.2 GA, which includes support for new data types ('tinyint', 'smallint', 'time', and 'date') and user defined function/aggregate (UDF/UDA) schema metadata. Additionally, the release introduces the whitelist load balancing policy, a streamlined schema metadata API, and several internal improvements. The driver now supports Apache Cassandra 2.2 and will focus on supporting Apache Cassandra 3.0 in the next release. Feedback is welcome to help prioritize future features.

Early stage projects often need to design and test a table that will scale on specific hardware configurations. Cassandra-stress is a tool that enables users to benchmark their data model without the app, providing valuable insights into performance and scalability. However, using cassandra-stress requires understanding of statistical concepts and CQL syntax. The CassandraDataModeler aims to simplify this task by allowing users to create user profile-driven cassandra-stress tests in minutes. Key factors to consider when designing a data model include partition key control, preventing hotspots, and ensuring that partitions are smaller than 100MBs and contain less than 100,000 cells. Additionally, understanding the size, population, and cluster distributions is crucial for creating realistic stress profiles. By using tools like htop and dstat during benchmarking, users can identify bottlenecks and optimize their Cassandra clusters.

The text discusses using map collections in DSE Search with Apache Solr for indexing. It explains that every key in the map must be prefixed with the name of the collection. The same methodology can be used in many ETL workloads. However, dynamic fields should not store large amounts of data as they create significant heap pressure due to their memory footprint. The text also mentions how DSP-5373 made it easier to index a table with a single API call and that the combination of cassandra-loader, cassandra-unloader, and some sed - awk magic can be used as a quick and dirty ETL tool.

This blog post discusses how to use Tuple and User-Defined Types (UDTs) with Cassandra version 4.8 and later. It provides a step-by-step guide on creating CQL tables, inserting data, and querying UDT and Tuple fields using the {!tuple} query parser. The post also covers how to handle collections of UDTS/Tuple, nested UDTS or Tuples, and dynamic fields. Additionally, it explains how to modify existing schemas by removing unwanted fields. Finally, the post highlights two limitations when using UDTs and Tuples in DSE Search: they are read/written as a single block, not on a per-field basis, and there is no support for updating individual fields within a Tuple or UDT.

The text discusses the use of Gremlin, a graph traversal language, in querying graph databases. It introduces "SQL2Gremlin" (http://sql2gremlin.com), a website that demonstrates how to produce similar results using SQL and Gremlin. The author provides three examples from the site: a simple filter-based query, a traversal of relationships without joins, and a recommendation query. The text also mentions DataStax's commitment to graph computing through its acquisition of Aurelius and its involvement in the Apache TinkerPop project.

Online fraud is a significant problem for businesses, with stolen credit cards being the primary contributor. DataStax Enterprise can help reduce exposure to fraud by handling millions of low-latency transactions per second, allowing real-time processing to prevent fraudulent transactions before they occur. The platform offers fault tolerance and high availability, ensuring continuous operation even during server failures or network issues. It also supports fast data reads and writes, enabling efficient transaction processing for credit card users. DataStax Enterprise integrates with advanced security features and partners like Vormetric to help organizations meet PCI-DSS requirements.

This post discusses using geospatial search in DataStax Enterprise (DSE) Search and with Apache Spark™ as part of DSE Analytics. It highlights the advantage of DSE allowing users to have one version of their data for both real-time access and specialized search queries, eliminating the need for ETL processes. An example project is provided on GitHub demonstrating how to load all postcodes in the UK with a longitude and latitude of their location, and how to query this data using geospatial searches. The post also explains how to connect to DSE data through Apache Spark using various methods such as CassandraTable method, CassandraConnector class, and Spark SQL. For more information, readers are encouraged to visit the DataStax Academy for tutorials, demos, and self-paced training courses.