Time-series databases pose great challenges for analyses at extreme scale, due to the velocity and volume of the data being ingested from multiple sources. Time is a first-class citizen of these datasets, and most workloads focus on aggregating or downsampling across time. It is important for ingestion, storage and processing of time-series data to be performed as efficiently and inexpensively as possible.
TileDB’s cloud-native, multi-dimensional array format considers time as a natively indexable dimension, and allows you to rapidly slice and aggregate massive quantities of time-series data. TileDB is up to 50 times faster and 10 times smaller in storage footprint than popular special-purpose time-series databases.
TileDB provides built-in time-series indexing, with support for NumPy and R datetime objects at any level of precision. Slice & dice massive datasets like historical stock tickers or user activity data with speed and at scale. Combine timestamps with other sources, like imagery, to build 3D and 4D time-series cubes for advanced geospatial and change detection use cases. Enjoy secure governance for collaboration and a 100% serverless architecture for scale, drastically reducing your TCO.