New features in Microstrategy 10.1

With the release of MSTR new version 10 and 10.1, there are various enhancement made by MicroStrategy corporation. Here are the few notes that BI people might be looking for.

First thing that everyone looking for is our architectural enhancement .So that MSTR will processed data more efficiently . Below are the major change in the architecture.

Intelligence Server  Architecture Enhancements

• Parallel Partitioned In-Memory Analytics  or PRIME
• Larger Data Volumes with Parallel Partitioned In-Memory Cubes
• Auto Unload Caches
• ROLAP SQL Engine: Enhanced SQL Global Optimization
• SAP BW Data Access Enhancement
• ODBC and JDBC Drivers

§  Parallel Partitioned  In-Memory Analytics  or PRIME :  It gives us the flexibility to  create in-memory cubes through Multi-Table Data Import. Prior to MicroStrategy 10, cube publishing was only possible using OLAP services. While that option is still available in MicroStrategy 10, Now we  have enhanced option, By using OLAP Services, the entire warehouse data that is published into a cube consists of a large, single table/Warehouse can be partitioned and the final cube can have several semi-DE normalized (i.e., integrated smaller tables into one big table) tables.

§  Larger Data Volumes with Parallel Partitioned In-Memory Cubes : It allows datasets and its individual tables to be partitioned across multiple cores. Each partition can have, up to 2 billion rows and it can scale linearly to thousands of partitions. Queries are now executed in parallel across multiple cores. With this features we can achieve higher throughput with parallel data fetching rates /data scalability with analyse more than 2 Billion rows and obviously faster response time since Analytical engine process to be done in parallel.

Comparison between In-Memory Vs Parallel Partitioned In-memory Cube

The below images shows how the both the technique enhance the performance .

1. Higher Throughput: This is because of the parallel ODBC data fetch with Prime, while it was single thread in 9.4.1
2. Higher Data Scalability: this is due to data partitioning available with Prime, hence we can now analyze more than 2B rows in a single cube
3. Faster Response Time: Prime enables analytical engine processes to be done in parallel and thus better performance. Also helps in post manipulations, for instance: making selections, filtering etc.

§  Auto Unload Caches : The Auto Unload Caches feature is meant to scale the number of projects that can be loaded on an Intelligence Server by proactively unloading caches of inactive projects. Typically report caches and Intelligent Cubes, consume the greatest amount of memory of all Intelligence Server subsystems. As a result, loaded projects, even in an idle state, consume Intelligence Server resources. However, with the Auto Unload Cache feature, without unloading a project , you can unload a project’s caches and Intelligent Cubes from memory whenever the project has been idle for longer than some specified threshold.  It is useful in environment that have  many projects that are sparsely used.

 

§  ROLAP SQL Engine: Enhanced SQL Global Optimization :  MSTR 10.1 introduces a SQL optimization features,  that enables SQL query passed get merged that hit the same warehouse fact tables . It reduces the no. of times that a report query hits a fact table. When multiple passes hit the same fact table ,this create a “Compiled table “ by combining the lookup tables of multiple passes then join the compiled table to fact table once.

 

§  SAP BW Data Access Enhancement : Using MSTR 10, we can connect to SAP Business warehouse instances and Support of Kerberos pass-through authentication.

 

§  ODBC and JDBC Drivers : Previously, MicroStrategy Analytics Platform only supported 32-bit ODBC drivers, with the exception of 64-bit ODBC drivers for HP-UX. Now MSTR 10 will no longer support 32-bit ODBC connections. We have now 64-bit supported driver along with third party ODBC  drivers.

Along with ODBC,  now MSTR also have JDBC drivers such as Presto, Hbase etc. These drivers can be used in an extension capacity to support databases/data sources that do not have the options to use ODBC.