Understanding how data is stored

All tabular data brought into Domo via connectors, Workbench, or APIs is stored in Vault. Vault is a persistent storage layer that feeds both the in-memory caching engine Adrenaline and the cache optimizer Tundra. Figure 3.1 illustrates these relationships:

Figure 3.1 – Persistent versus in-memory storage

Vault resides on major cloud infrastructure file services from either AWS or Microsoft Azure. The default cloud service is AWS unless Azure is specifically requested. The physical data is stored and versioned in Vault, regardless of the intake mode (Append or Replace) on the dataset. In Replace mode, previous data versions are no longer queryable from the UI, but they are still stored as an older version of the data. In Append mode, all data versions are queryable because they are linked together as a single entity as illustrated in Figure 3.2:

Figure 3.2 – Dataset version chaining...