Cassandra storage engine

This page contains documentation for the Cassandra storage engine which is a work in progress.

Data mapping

A table in MariaDB represents a column family in Cassandra. The table must follow this pattern:

create table cassandra_tbl      -- Table name can be chosen at will
(
  rowkey  type PRIMARY KEY,     -- represents Column Family's row key. Primary key
                                -- must be defined over the column.

  column1 type,                 -- Names of columns must match the names of CQL's 
  column2 type,                 -- columns; datatypes must also "match", see below
                                -- for what this means
  ... 
  dyn blob DYNAMIC_COLUMN_STORAGE=yes -- It is not mandatory but only one
                                      -- blob column with any name
                                      -- to collect all columns of the record
                                      -- which are not mentioned in
                                      -- MariaDB table definition
) engine=cassandra
  keyspace= 'cassandra_key_space'
  column_family='column_family_name'
  thrift_host='192.168.1.0'           -- can also be set globally with 
                                      -- @@cassandra_default_host variable.
  thrift_port='9160'   -- optional, use only if non-default.

The MariaDB table is a view of a Column Family in Cassandra. That is,

• Creating a table will not create a Column Family. The Column Family must exist in Cassandra before the MariaDB table can be created.
• Dropping the table will not drop the Column Family.

Mapping for columns

1. The row key. The Cassandra API does not consider it to be a column, but CQL and MariaDB do.
   • If the Column Family defines a name for the rowkey (always true when using CQL), then the name of the primary key column in MariaDB must match it.
   • If the Column Family doesn't define a name for the row key, MariaDB's column must be named "rowkey".

2. Columns that are defined as part of Cassandra's "static column family" (this includes columns that one can define with CQL). These columns should have their counterparts in MariaDB. A static column named 'foo' in Cassandra must have a counterpart named 'foo' in MariaDB. The types must also match, they are covered below.

3. "Ad-hoc" columns that can be encountered in individual rows and static columns which are not part of mariaDB table definition mapped into one blob column with Dynamic Columns content. Such column have to be blob and marked with boolean option DYNAMIC_COLUMN_STORAGE with value meaning true (yes/on/1). If its name match Cassandra's column the Cassandra column also will be included into the blob. If such column absent all columns which are not listed in the MariaDB table definition will not be read and can't be changed.

4. Cassandra's SuperColumns are not supported, and we have no plans to support them.

Mapping for datatypes

There is no direct 1-to-1 mapping between Cassandra's datatypes and MySQL datatypes. Also, Cassandra's size limitations are often more relaxed than MySQL/MariaDB's, for example, Cassandra's limit on rowkey length is about 2G, while MySQL limits unique key length to about 1.5Kb.

Mapping between Cassandra types and MySQL types are as follows:

Command mapping

INSERT

Cassandra doesn't provide any practical way to make INSERT different from UPDATE. Therefore, INSERT works as INSERT-or-UPDATE, it will overwrite the data, if necessary.

INSERT ... SELECT and multi-line INSERT will try to write data in batches. Batch size is controlled by the @@cassandra_insert_batch_size system variable, which specifies the max. batch size in columns.

The status variables Cassandra_row_inserts and Cassandra_row_insert_batches allow one to see whether inserts are actually batched.

UPDATE

UPDATE works like one would expect SQL's UPDATE command to work (e.g. changing primary key value will result in old record being deleted and new record being inserted)

DELETE

• DELETE FROM cassandra_table maps to truncate(column_family) call.

• The DELETE with WHERE clause will do per-row deletions.

SELECT

Generally, all SELECT statement work like one expects SQL to work. Conditions in form primary_key=... allow the server to construct query plans which access Cassandra's rows with key lookups.

Full table scan

Full table scans are preformed in a memory-efficient way. Cassandra SE performs a full table scan as a series of batches, each of which reads not more than @@cassandra_rnd_batch_size records.

Batched Key Access support

Cassandra supports Batched Key Access in no-association mode. This means that it requires the SQL layer to do hashing, which means the following settings are required:

• optimizer_switch='join_cache_hashed=on'
• join_cache_level=7|8

Cassandra SE is currently unable to make use of space in the join buffer (the one whose size is controlled by @@join_buffer_size). Instead, it will limit read batches to reading not more than @@cassandra_multiget_batch_size at a time, and memory will be allocated on the heap.

Note that the @@join_buffer_size buffer is still needed by the SQL layer, so its value should still be increased if you want to read in big batches.

It is possible to track the number of read batches, how many keys were looked-up, and how many results were produced with these status variables:

Variable_name	Value
Cassandra_multiget_reads	0
Cassandra_multiget_keys_scanned	0
Cassandra_multiget_rows_read	0

How to try it out

See Building Cassandra Storage Engine.

See also

• MDEV-431 - jira task for Cassandra SE works
• cassandra-storage-engine-future-plans
• cassandra-storage-engine-use-example
• cassandra-storage-engine-issues
• hbase-storage-engine