Some of the most common questions from customers are about their analytics administration. Often regarded as house-keeping, administration failures can nevertheless cause real pain.
The typical day-to-day tasks of analytics administration include checking availability and health of the analytics platform components, allocating proper resources for users like memory, file systems, databased and also auditing user activity. The scope of this work is fundamentally defined by the requirements for any analytics solution to perform effectively i.e. software and underlying hardware to be architected optimally, managed metadata and user access and configuration to support business needs.
IoT scale and realtime demands
With the advance of IoT in many organisations, it is no surprise that some of the administration related questions have been triggered by the nature and scale of IoT data. Thousands of sensors streaming data via gateways will still have to be made available for models. The number of models to interrogate this data can be in the thousands.
Each of the normal administration tasks stands to be amplified when handling event streams and associated analytics models. With the realtime perspective, all these tasks become crucial and more important. Using right tool for checking resources usage and correct resource allocation is the key. High availability, fault tolerance and resource elasticity will all determine success of the analytics production run.
Increasingly diverse stakeholders
The other area of significant change is stakeholder management. For the most part, analytics administrators have been working with Information Technology (IT) support teams to implement policies, securing resources and troubleshoot issues. With IoT, some of the infrastructure will belong to Operations Technology (OT). This means there are two distinct resource pools that need to be liaised with for typical operational matters.
Analytics administrators also typically work with Data Scientists and business users who are looking for new environment, better performance and information about usage and audits. With IoT, the user community is much more diverse. The ‘citizen data scientist’ tribe will be further empowered by IoT and will need to be served in a slightly different way.
Re-defining availability and deadlines
Analytics administrators are accustomed to deadlines, especially if there is also the need to provide high level of platform availability during peak times. Users’ activities are normally not at the same level through months and year; For example the end of the months is often hot for generating reports and analysis. With IoT, these deadlines will be a constant feature. Insights need to be generated constantly, in realtime. Administrators will be critical to ensuring the smooth running of the analytics factory.
If you want to know more about how the Internet of Things delivers fast-moving data from sensors and devices, read the white paper Understanding Data Streams in IoT