The goals of a retention strategy for IoT analytics are twofold:

• Maintain Value: Advanced modeling techniques, such as deep learning, need lots of history to maximize prediction effectiveness. It is also difficult to know ahead of time which fields will be valuable for a future unknown project. The traditional data retention strategies of storing records for a fixed period of time and then deleting the full dataset could result in lost profitable opportunities. A different mindset is needed.
• Minimize Costs: IoT data can get big quickly. Even with cloud-based HDFS storage, costs can get large. Keeping everything forever could easily cost more than the value it provides. The more accessible data is, the more costly it becomes. There should be some compromises to keep your costs low.