Implementing a Search and Analytics Engine in a Ten-Year-old Mission

The analysis and querying of spacecraft telemetry are crucial for mission operations, providing valuable insights into the performance and health of the spacecraft. For the Mars Science Laboratory (MSL), telemetry has accumulated over 11 years on the same databases and negatively impacts the efficiency of querying and visualizing the data.To mitigate these issues, MSL proposed a third-party search and analytics engine called Elasticsearch (ES). This engine was implemented successfully for mission operations and continues to be used for strategic and planning purposes. However, there are other systems in the mission that generate telemetry: the testbeds. MSL utilizes two of these test venues on Earth at the Jet Propulsion Laboratory (JPL) to simulate various situations as it would occur in flight. Although the testbed hardware is a close replica of what is on-board Curiosity on Mars, data is transferred more reliably in the testbed due to the physi- cal location of the operators and receiving computers. This results in a much greater volume of testbed data than flight data, which remained in a Advanced Multi-Mission Operations System (AMMOS) Mission Data Processing and Control System (AMPCS) database and had to be queried through a series of tools called Chill. Chill, a wrapper around Structured Query Language (SQL) commands, allows users to query the MySQL AMPCS database. As more testbed sessions occurred over the years, querying testbed telemetry became impractical for longer time ranges, making the MSL testbed team resort to Elasticsearch as their solution.A current generalized Elasticsearch pipeline on MSL includes a data store (cloud or relational database), code utilizing the ES Application Programming Interface (API) to collect and ingest the data from the store, and a front-end interface to query and view the ES data. However, during development, the testbed team encountered limitations while ingesting millions of data points. By finding solutions to these limitations, the feasibility to ingest a given amount of data can be determined by available disk space, the amount of data points to ingest, and time constraints. As a result, the testbed team was able to significantly cut down on querying times, ease flight trou- bleshooting, generate insightful testbed trending information, and adapt multi-mission tools.As future missions consider newer technologies that offer simi- lar services to Elasticsearch, it is pertinent to quantify the time and effort it takes to implement such a technology, in addition to the cost. Although Elasticsearch was innovative and brought many benefits to MSL, findings from this experience highlight the importance of knowing the boundaries of such technologies prior to official implementation.