SSTableLoader is a command-line utility for bulk-loading SSTables from Cassandra or ScyllaDB into a Scylla cluster. It offers a straightforward migration path when the source data already exists as SSTables. section 12.2.1 section 12.2.2
Use SSTableLoader by preparing the data and destination in this order:
- Create the destination schema before loading; the tool assumes the schema for the SSTables already exists. section 12.2.2
- Arrange the files in a directory resembling Scylla's data layout, with keyspace and table subdirectories. section 12.2.2
- Point SSTableLoader at the directory; it scans the SSTables, converts their contents into CQL statements, and executes them on the destination cluster. section 12.2.2
Restoring a reviews table
After snapshotting the reviews keyspace, move the SSTables for the articles table into reviews/articles. Point SSTableLoader at that directory to load the table into Scylla. section 12.2.2
SSTableLoader is easy to set up but narrow in scope. It cannot load from non-Cassandra or non-Scylla sources, lacks checkpointing, and can be parallelized only per table. During an online migration, pair it with application dual-writing and validate correctness with dual-reading. section 12.2.5 section 12.3
Figure 12.5 The SSTableLoader, as its name suggests, migrates data by reading a set of SSTables and inserting their contents into Scylla.
Shows the loading flow: SSTableLoader reads SSTables and inserts their contents into Scylla. section 12.2.2
