• Running on shared-nothing clusters of commodity hardware
• In-memory database
• SQL support
• ACID
• Linear Scalability
• Released as an Open Source software

Actually, there have already been some OLTP databases running on shared-nothing clusters. However, they cannot take advantage from the scalability of shared-nothing architecture due to their implementation’s natures, such as complex distributed locking and commit protocols [1]. In addition, according to [3], traditional RDBMSs have four overhead components, which are logging, locking, latching, and buffer management. However, M. Stonebraker claims that VoltDB eliminated these legacy overheads.

Among many features, especially I have interest in its linear scalability with ACID and performance. It is meaningful in that today’s web applications have another alternative to NoSQL data stores. Although VoltDB is under heavy development, the above features and the next benchmark result show its promising.

• Key-Value Benchmark (VoltDB versus Cassandra)

Cassandra is a remarkable key-value store and an open source project developed by apache committers. Now, it is well known as the most performant one in existing NoSQL stores. According to this benchmark result, however, in all cases VoltDB dominates Cassandra although the fairness of experiments is controversial.

• VoltDB Roadmap

It’s future plan is also expected. I wonder how much attention VoltDB will be getting from communities and industrials.

See Also:

1. Data Management in the Cloud: Limitations and Opportunities
2. Comparing VoltDB vs Postgresql
3. OLTP through the looking glass, and what we found there, ACM SIGMOD 2008
4. http://voltdb.com/product
5. H-Store: A Next Generation OLTP DBMS