Native workspaces
Cassandra and ScyllaDB
Apache Cassandra and ScyllaDB open in a CQL workspace. It browses a table one partition at a time, pages the way the driver does, runs CQL, and stages inserts, updates and deletes as the exact statements they will run.
Where to find it
| Action | How |
|---|---|
| Create a connection | New connection, ⌘N (Ctrl+N), then Apache Cassandra or ScyllaDB |
| Open the workspace | Connect on a saved connection |
| Browse keyspaces and tables | CQL catalog on the left, with Show system keyspaces for the system ones |
| Read a partition | Select a table, enter each partition key, then Load partition |
| Run CQL | CQL editor, then Run |
| Stage a change | Staged CQL mutations, then Stage and validate |
| Switch connection | ⌘⇧K (Ctrl+Shift+K) |
The catalog sits on the left, the editor and its history across the top, and the results beside the staged mutations below.
Connecting
Choose Apache Cassandra or ScyllaDB in New connection. Enter Contact points as comma-separated seed hosts, the CQL port, the Local datacenter and an optional Default keyspace, with a Username and Password when the cluster requires them.
TLS verification is Disabled or Verify certificate and hostname, with optional PEM files for a CA, a client certificate and a key. Consistency and Serial consistency set the defaults for the connection.
Browsing by partition
Select a table and an input appears above the results for each partition key column. Rows load only when every partition key has a value and you choose Load partition, because the workspace never adds ALLOW FILTERING, token scans or OFFSET.
Each value is a CQL literal, so text needs single quotes. Results arrive 100 rows at a time, and Next driver page continues from the paging state the driver returned.
Values
Cells cross from the driver to the window as tagged values, so decimal and varint keep every digit and nothing is rounded. Sets, maps, tuples and user-defined types keep their structure, a blob shows as hexadecimal and a timestamp as milliseconds. Hover a cell to read it as a CQL literal you can paste into the editor.
CQL
The CQL editor runs one statement with Run, at the consistency chosen beside it: ONE, LOCAL_QUORUM, QUORUM or ALL. Named markers take their values as JSON in Prepared values. History keeps the last 50 distinct statements on this computer. A server error, such as a refusal that asks for ALLOW FILTERING, appears above the results with the server's message.
Editing
Choose Insert, Update or Delete, write one column = CQL literal line per value and, for an update or delete, per key column, then choose Stage and validate. Inserts need every partition key, updates and deletes need the complete primary key, and TTL seconds adds USING TTL to an insert or update.
A condition turns an update or delete into a lightweight transaction with IF, and any text in the insert condition box adds IF NOT EXISTS. Apply staged runs the listed statements in order, and because CQL has no rollback across statements, a failure reports how many earlier statements remain applied.
Good to know
- ScyllaDB is its own provider with a
scylladb://URL, and shares this workspace with Cassandra. - A read-only connection refuses write and schema CQL and staged statements before they reach the cluster.
Cancelstops waiting for a request, but a coordinator that already accepted it may still complete it.- There is no row inspector, schema designer, diagram, snapshot, import or export here. CQL tables have no foreign keys to draw, and the SQL dump and CSV tools do not describe CQL data.