Overview


What Is LeanXcale

LeanXcale is a real-time big data platform that can scale in any of the three Vs of Big Data (Volume, Velocity and Variety). It can scale in Volume (to 100s of terabytes), in data velocity (to million transactions per second, and millions of events per second) and even in data variety (structured, non-structured, key-value, streaming). Let us examine each of the facets separately.
Ultra-Scalable Full ACID Full SQL Database

Ultra-Scalable Full ACID Full SQL Database

LeanXcale is an ultra-scalable full ACID full SQL database (OLTP database). The main distinctive feature lies in that it can scale from 1 node to 100s of nodes and able to process very large update transaction rates. It is based on a radically new approach to transactional processing that is able to provide transactional data consistency while being able to scale out to large numbers. The scalability is fully transparent. Syntactically the database is accessed by means of a standard JDBC driver. Semantically it exhibits the same behavior as any full ACID database. With LeanXcale sharding will never be needed anymore to scale your database.

Blending OLTP and OLAP functionalities

Real-Time Analytics OLTP+OLAP

Real-Time Analytics OLTP+OLAP

LeanXcale is equipped with an OLAP engine. The OLAP engine parallelizes queries over multiple nodes to enable to answer heavy analytical queries in online response times. The OLAP engine works over the operational data, what is possible thanks to the ultra-scalability of the transactional processing. This combination enables LeanXcale to deliver real-time analytics capabilities. It can also be used to reduce the number of ETLs. Operational data can be kept in LeanXcale that can be also queried with analytical queries, avoiding the traditional ETLs from operational databases to data warehouses. ETLs are estimated to be 80% of doing business analytics what is a total nonsense.

Operational Data Lakes

Operational Data Lakes

Operational Data Lakes

Data lakes are being widely adopted by enterprises. However, the use of the data lakes brings some disadvantages. Some technologies such as map-reduce and evolutions are programmatic what is expensive in terms of development. Other technologies enable to query data in subsets of SQL such as Hive or Impala. However, these technologies can only access the data in the data lake what is insufficient in many cases. With LeanXcale by defining the metadata and parsing of Hadoop files, they become read-only SQL tables. However, all the operational data stored in LeanXcale is also available enabling to query both unstructured and structured data and historic information in Hadoop data lakes and operational data in an OLTP system.

Polyglot Data Management

Polyglot Data Management

Polyglot Data Management

LeanXcale is integrating multiple data management technologies to empower it with persistence polyglot capabilities (term introduced by Fowler in his book “NoSQL Distilled”). LeanXcale is able to work over structured, unstructured and semi-structured data. Structured data is handled via the SQL support. Unstructured data is supported by means of the Hadoop data lake integration just discussed. Semi-structured data is supported by means of providing a transactional key-value data store, more concretely, a transactional version of HBase with full ACID properties. In this way, LeanXcale can handle all kinds of stored data.Additionally, LeanXcale has been integrated with a scalable data streaming technology, Apache Storm, to enable to handle streaming data and correlate it with stored information in LeanXcale. This integration has been shaped as a Storm bolt that basically enables to write arbitrary SQL statements to be executed for each event. In this way, the even can be correlated with information stored in LeanXcale and can also store or update information in LeanXcale. What is more, it is also possible to materialize as SQL tables the output of Storm queries through a materialization bolt we have produced to enable some applications to read the output of Storm queries from SQL tables so they do not have to deal with the Storm data streaming API, but simply with SQL.


Comments are closed, but trackbacks and pingbacks are open.