Data Ingestion

Data ingestions are achieved via upsert batches with the following procedures:

1. Client contacts the server for schema resolution before preparing an upsert batches. This converts column names to column IDs.
2. Client contacts the server for auto-enum value resolution. This converts all string values of auto-enum columns to integers. In distributed setup all servers need to agree on the same schema and auto-enum dictionary.
3. Client prepares an upsert batch using the serialized format. Unlike a SQL insert, AresDB does not provide transaction atomicity and isolation higher than record level.