Big data and analytics are a major priority for defense organizations. Data analytics, in particular, can enhance understanding of operational readiness, improve real-time situational awareness, and swiftly produce “what-if?” scenarios. Information sources include data within the ERP, the large amounts of data progressively fed to ERPs by on-condition health management systems (which prolong the life of equipment by monitoring its condition and indicating the need for maintenance or repairs), plus external data feeds such as RFID, GPS and geospatial. Most legacy ERPs and their underlying database technologies cannot support the in- and outflows of data (structured, unstructured, and geospatial) nor the required speed of processing.
Should an ERP system support either or both of these options?
|Big data||Fast data|
|Data at rest||Data in motion|
|Mine large static files||Process events at high velocity|
|Analyze later||Analyze now|
|On disk||In memory|
|Hadoop, MapReduce, NoSql technologies||Event processing to direct analysis|
Source: KPMG International, 2016.
Current datamarts and data warehouses contain big data that has been batched from ERP systems; an approach that is still relevant (subject to batch download times and data latency). However, to achieve fast data, defense forces must review their entire architecture and move towards in-memory capabilities (using Random Access Memory rather than a hard disk) and real-time event processing.
SAP HANA and Oracle Exalytics – which are gaining in popularity as add-on, fast data functionality for defense organizations – both provide extreme, in-memory analytics for Business Intelligence and Enterprise Performance Management (EPM) applications. Given that the design, and intended use, of these tools differs from previous analytics solutions, defense forces will have to determine whether they are suitable for battlefield use.
Relational databases can now offer ‘tenanted capabilities’ allowing retention of multiple security levels within a single database (with one software application serving multiple customers, or ‘tenants’). Structured, unstructured and geospatial data can also be fluidly stored and processed within single and multiple databases. Furthermore, the evolution of database and hardware capability enables concurrent access to row and column data, to allow simultaneous transactions and reporting within a single system.
Smart database technologies, where data cell locations are stored in memory (allowing direct access instead of an index grid search), combined with in memory capabilities, all positively aid data analysis. Emerging technologies such as binary object streaming, data release acceleration and integrated hardware/database/chip and infiniband design similarly open up analytical opportunities, as do continuous query language (CQL) and joint management of non-relational language (NOSQL) and structured query language (SQL).
The key to harnessing the power of data lies with linking the data to the needs of the user and understanding what capabilities the system needs to bring so they can innovate.