# Big Table - [Main page](https://cloud.google.com/bigtable) - A fully managed, scalable NoSQL database service for large analytical and operational workloads with up to 99.999% availability. # Tech thinks to remember - Wide column database - Multidimensional Sparse Array - Scales linearly - Read/Write is atomic at row level - Hbase API (Hadoop) - Consistency - eventually consistent - but you can use read-your-write consistency - strong consistency - Row key is like primary table but BigTable does not support join - the row key value determine where the data is written - Columns families - set of related columns in a table - product/name/size - 1000 tables per instance - Better to keep all data in one table - Empty cell consume no space ## Recommendations - # of column family < 200 - Keep all value in a row below 100MB - Keep cell values under 10MB - Keep row key below 4k ## Limits - Single value < 100MB - A row < 256MB ## Concepts about data - Metadata is saved on a node and data is saved in the Colossus file system ## Use cases - Petabyte scale - Low latency writes - Key-based reads - Analytic and ML - Time series applications