Apache Accumulo

Apache Accumulo is a mashup of various technologies, from Google's BigTable, to Apache's Hadoop, Thrift and Zookeeper.

Compared to Google's BigTable system, Accumulo features a few improvements of its own.

These include table cell-based access restrictions, a server-side system for managing key-value pairs at desired times and in optimum conditions, and lots of client APIs.

The database is certainly not for running your every day websites and is targeting for cloud-computing environments where developers need to handle humongous amounts of information.

What is new in this release:

• Use of Hadoop CredentialProviders
• Write-Ahead Log sync performance
• Minor-Compactions not aggressive enough
• Write-Ahead log sync implementation
• HeapIterator optimization

What is new in version 1.6.2:

• Use of Hadoop CredentialProviders
• Write-Ahead Log sync performance
• Minor-Compactions not aggressive enough
• Write-Ahead log sync implementation
• HeapIterator optimization

What is new in version 1.6.0:

• Service IP addresses
• Multiple volume support
• Table namespaces
• Pluggable compaction strategies
• Conditional mutations
• Locality groups in memory
• Size-based constraint on new tables

What is new in version 1.4.1:

• Optionally monitor swappiness on every server.
• Support running on-top of Kerberos-enabled HDFS.
• Provide method for gathering system stats to API.

What is new in version 1.4.0:

• Tablet merging
• Efficient deletion of row range
• Compaction of row range
• Table cloning
• FATE: Fault Tolerant Executor. Used to make table operation survive master restart.
• Concurrent table operation execute correctly
• Bulk load is now done by master and tablet servers and uses FATE to survive server restarts.
• Multi-level RFile index
• Merging minor compactions
• Logical time for bulk import