Without a retention policy, IoT storage grows forever and the bill with it. A typical scheme keeps raw readings for a short period for troubleshooting, minute or hour averages for a longer one for trend analysis, and daily summaries indefinitely. A time-series database usually enforces this automatically through retention rules and continuous downsampling.
Retention is not only a cost question. Privacy law may require that personal data, including location traces and energy profiles that identify a household, is not kept longer than needed, while contracts or safety regulations may require that certain records are kept for years. Writing those requirements down per data class is part of data governance, and it is easier to do before the first terabyte than after.
Part of the IoT Glossary. Spotted an error or a missing term? Tell us.