Data modelling and design
As data volumes rapidly increase, businesses become more globalised and technology options grow, a deep understanding of how to design and organise data to meet complex requirements is increasingly becoming a critical success factor in data projects.
Many projects start with large upfront analysis and data acquisition phases without fully understanding the business problems they are trying to solve and the features their data will need to support in order to deliver efficient decision making. Consequently data projects frequently get to the point of no return before key feature gaps are discovered, resulting in significant additional investment and delivery delays.
Rather than design from first principles, Kinaesis consultants are able to draw upon tried and tested data design patterns from our proven data management best practices. These patterns are then applied to accelerate client delivery and mitigate the risk of developing solutions that are not fit for purpose.
Our client had already started a multi-year strategic programme to centralise daily P&L and Balance Sheet production, however did not have the experience in-house to implement the demanding global reporting requirements.
At the forefront of these business needs were integrated workflow and the ability to report data against the temporal nature of the business workflow whilst maintaining performance of their reporting requirement across both base and adjusted views of data.
Prior solutions had taken a snapshot approach to this problem, leading to internal reconciliations between snapshots and increasing levels of data redundancy.
After reviewing client requirements and performing several rapid iterations to prove the design against the core business event workflow and month end adjustment requirements, we implemented a high performance (insert only) bi-temporal data model.
In addition, ensuring the data model functionally supported all known and many unknown reporting requirements, our business event labeling pattern was used to provide the reporting layer with point-in-time query capabilities.
For the first time, Product Controllers and Front-Office were able to select a single state driven view of their data where one example would be to display reports for a specified Business Unit when all Traders had approved their P&L.
Using the same underlying temporal capabilities, users were also able, for the first time, to run reports to give them progressive views (and regressive, if required) of P&L through the 3 day post month-end adjustment process. These advanced temporal capabilities have eliminated the need for users to allocate resource to reconcile different versions of data corresponding with different business events.