MariaDB starting with 10.0.4

The Spider storage engine was introduced in MariaDB 10.0.4

Not complete yet - still being updated

F(*) Federation only , P(*)partioning only . Spider column is for SpiderForMySQL found on the Spider web sIte.

FeatureSpider10.0
Clustering and High Availability
Commit, Rollback transactions on multiple backendYesYes
Multiplexing to a number of replicas using xa protocol 2PCYesYes
Split brain resolution based on a majority decision, failed node is remove from the list of replicasYesYes
Enable a failed backend to re enter the cluster transparentlyNoNo
Synchronize DDL to backend, table modification, schema changesNoNo
Synchronize DDL to other SpiderNoNo
GTID tracking per table on XA errorNoYes
Transparent partitioningNoNo
Covered by generic SQL test caseYesYes
Heterogenous Backends
MariaDB and MySQL database backendYesYes
Oracle database backend, if build from source against the client library 'ORACLE_HOME'YesYes
Local table attachmentYesYes
Performance
Index Condition PushdownNoNo
Engine Condition PushdownYesYes
Concurrent backend scanYesNo
Concurrent partition scanYesNo
Batched key accessYesYes
Block hash joinNoYes
HANDLER backend propagationYesYes
HANDLER backend translation from SQLYesYes
HANDLER OPEN cache per connectionNoYes
HANDLER use prepared statementNoYes
HANDLER_SOCKET protocol backend propagationYesYes
HANDLER_SOCKET backend translation from SQLNoNo
Map reduce for ORDER BY ... LIMITYesYes
Map reduce for MAX & MIN & SUMYesYes
Map reduce for some GROUP BYYesYes
Batch multiple WRITES in auto commit to reduce network round tripYesYes
Relaxing backend consistencyYesYes
Execution Control
Configuration at table and partition level, settings can change per data collectionYesYes
Configurable empty result set on errors. For API that does not have transactions replayYesYes
Query Cache tuning per table of the on remote backendYesYes
Index Hint per table imposed on remote backendYesYes
SSL connections to remote backend connectionsYesYes
Table definition discovery from remote backendYesF(*)
Direct SQL execution to backend via UDFYesYes
Table re synchronization between backends via UDFYesYes
Maintain Index and Table Statistics of remote backendsYesYes
Can use Independent Index and Table StatisticsNoYes
Maintain local or remote table incrementsYesYes
LOAD DATA INFILE translate to bulk insertingYesYes
Performance Schema ProbesYesYes
Load Balance Reads to replicate weight controlYesYes
Fine tuning tcp timeout, connections retryYesYes

Comments

Comments loading...