MotionQL

More databases

Cassandra and ScyllaDB

Keyspaces and tables, rows, the CQL console, schema changes and the cluster panel.

MotionQL works with Apache Cassandra 3, 4 and 5, ScyllaDB and DataStax Enterprise over the CQL native protocol. The server is detected when you connect. DataStax Astra DB can be reached with its secure connect bundle, but Astra support is cloud-only and untested: it has only been checked against mocks, never against a live Astra database.

Connecting

  • Click + under Apache Cassandra / ScyllaDB in the sidebar.
  • Cassandra / ScyllaDB: enter one or more contact points (comma separated; host:port overrides the port), the port (9042 by default) and the local data center, which nodetool status shows. Requests go to the nodes of that data center. Add a user name and password for PasswordAuthenticator. Pick a keyspace to USE on connect and a default consistency (LOCAL_ONE when none is set).
  • TLS turns on encryption. You can choose a CA, a client certificate and a key (PEM files, read when you connect), and you can turn off certificate verification for local servers only.
  • Astra DB (bundle): choose the secure-connect-<database>.zip file, then enter token as the client id and the AstraCS:… application token as the secret.
  • Test connection shows the server and its version, the cluster, the data center, the number of nodes and the protocol version. When the test fails, Ask AI why it failed explains the error and can offer a fix, such as the port or TLS (see AI help). Passwords and tokens are kept encrypted.
  • Read-only connection refuses every write: rows, CQL INSERT / UPDATE / DELETE / BATCH, DDL and TRUNCATE. SELECTs still work. Production mode shows a colour bar, confirms every write, and asks you to type the name before you drop or truncate anything. DROP and TRUNCATE always need the name typed. Every write and schema change is written to the audit log.

Browsing

  • Under a connection the sidebar lists the keyspaces, read from system_schema, with their replication (Simple, RF 1, or each data center's factor) and whether durable writes are on. System and virtual keyspaces are grouped apart.
  • A keyspace holds Tables, Materialized views, Indexes (SAI and custom index classes are named), User types, Functions and Aggregates. Expand a table to see its columns, marked as partition key (PK), clustering (CK, with ↓ for descending) or static.
  • Right-click an object for Show CREATE … and Drop …. On a keyspace you can also Create table… or Alter keyspace (replication)…; on a table, Alter table…, Create index…, Create materialized view… and Truncate….

Rows

  • Opening a table shows its Rows and Details. Rows are read in pages of 25 to 1000 using CQL paging state. The arrows go forward and back, and Show CQL shows the exact SELECT.
  • Add filter builds the WHERE clause. The builder knows the primary key. It says when a filter reads one partition. When a filter would need ALLOW FILTERING (a regular column without an index, or a clustering column without the columns before it), it warns you and explains why. The query only runs with ALLOW FILTERING when you tick Run with ALLOW FILTERING anyway.
  • A consistency picker sets the read consistency for the page.
  • Double-click a row to edit it. The primary key cannot change, and only the columns you changed are written. Add row inserts a row (an upsert, or IF NOT EXISTS as a lightweight transaction), with an optional TTL. Tick rows and Delete selected removes them by primary key. It asks first, and in production mode it asks you to type the table name.
  • CSV and JSON export every row that matches the current filter.

CQL console

  • Open it with New CQL console on a connection, Query in CQL on a table, or Mod+T on a Cassandra tab. It has highlighting and completion of keywords, types, functions, keyspaces, tables and columns.
  • Ctrl+Enter runs the statement under the cursor. If you select several statements, they run in order, each with its own result. You can set the keyspace, the consistency and the page size. More rows continues a large result.
  • Tracing turns on TRACING for each statement and shows the trace events: the activity, the node and the elapsed time, on a timeline.
  • Statements that use ALLOW FILTERING are flagged. Writes ask first in production mode and are refused on read-only connections. History is kept per connection.

Schema changes

  • Create keyspace… and Alter keyspace… set SimpleStrategy or NetworkTopologyStrategy, the factor for each data center, and durable writes.
  • The table designer sets the columns and their types, the partition key, the clustering columns and their order, static columns, compaction (size-tiered, leveled, or time window with its window), the default TTL, gc_grace_seconds and a comment. It previews the exact CQL first. Alter table adds and drops columns and changes the options. Dropping a column asks first.
  • Create index… creates a secondary index or an SAI index on a column, collection values, keys, entries or a FULL frozen value. Create materialized view… builds a view from a base table and its full primary key.
  • Drop (keyspace, table, view, index, type, function, aggregate) and Truncate ask you to type the name.

Cluster panel

Cluster, nodes and settings on a connection shows the nodes (address, status, data center, rack, version, tokens, host id and schema version), table sizes from system.size_estimates (it warns about large partitions), and, on Cassandra 4 and later, the settings, clients, thread pools and caches virtual tables.

AI help

When AI is allowed for the connection, Ask AI on a table or in the CQL console can:

  • write CQL from plain English;
  • explain or fix a statement;
  • rewrite a query that needs ALLOW FILTERING into one that reads one partition;
  • advise on modelling (a query table for each access pattern, partition size, clustering order).

It sends the table's schema, keys, indexes and size estimate, never rows. Statements it writes are checked before you see them. The check flags writes on read-only connections, destructive statements, ALLOW FILTERING, and reads that touch every partition. The statements are only put in the console, never run. The MCP server has read-only tools for open Cassandra connections: cassandra_list_tables, cassandra_describe_table, and cassandra_select (one SELECT of at most 20 rows; values are masked unless the connection allows them).

Something unclear or out of date? Write to support@motionql.com.