There are three elements to the Actian Analytics Platform: data integration, data preparation and what is sometimes referred to as the Extended Data Warehouse (XDW).
A traditional data warehousing environment has been considered to consist of a conventional data warehouse plus data marts. The key element that has changed with big data and The Internet of Things is the requirement for a discovery platform where large amounts of data can be explored for interesting insights but in a relatively inexpensive manner. In the case of the Actian Analytics Platform this would be its new Hadoop SQL edition offering, which is a combination of DataConnect for high performance connectivity, DataFlow for visual data science and analytics, and the implementation of Vector on HDFS (the storage system for Hadoop) via YARN. Vector is a column-based analytic database for single server implementations that owes its performance advantages (it currently holds the TPC-H benchmark record for the fastest non-clustered database solution) to the fact that it uses vector processing that exploits CPU parallelism. What Actian has now done is to implement this on top of HDFS so that it will run on low cost Hadoop clustered hardware. Initial benchmarks suggest that Vector in Hadoop is better than an order of magnitude faster than Impala running on Hadoop.
Once interesting nuggets of information are discovered the typical process is that relevant data will be loaded into a high performance analytics database for low latency analytic processing and operational business intelligence. Here Actian offers Vector, Matrix and SPARQLverse. Vector we have already discussed. Matrix (formerly ParAccel) is an MPP (massively parallel processing) column-based database. In technical terms the product’s biggest strength is its optimiser, which has a number of advanced capabilities (for example, for processing correlated sub-queries). It is worth noting that Matrix is what powers Amazon Redshift. SPARQLverse is an MPP-based graph database running on top of HDFS developed by SPARQL City (of which Actian is a major shareholder).
For data preparation, blending and enrichment, and visual discovery analytics and data science, the company offers Actian DataFlow (previously Pervasive DataRush), which is a massively parallel engine (it is extremely fast) that allows you to profile, cleanse and run other data preparation tasks, and/or execute data mining algorithms on the fly: that is, while loading data. It is worth commenting that the market for data preparation and analysing data while in motion is burgeoning and rapidly evolving into a sub-market of its own (several new and existing vendors are introducing markets in this area) but Actian got here first.
Finally, Actian DataConnect not only includes the data profiling and data quality capabilities that may be instantiated on Actian DataFlow but it also provides the data integration (ETL: extract, transform and load) requirements necessary to ingest and output data from and to both external and internal sources and targets. It is noteworthy that Pervasive, prior to its acquisition, had a history of developing connectors to a very wide range of sources and targets and, over the last few years in particular, it has focused on developing adapters to connect to cloud-based and SaaS (software as a service) providers. It has a particularly close relationship with Salesforce.com.
