Before the revolution in the database industry that saw the number of platforms go from 4 to around 130, we lived in a world where OLTP systems could do a few thousand ACID transactions per second with millisecond latency, but struggled to scale beyond that.
When it came to large volumes of data - such as the billions per day the telephone system produces – we not only struggled to produce responses in hours, but we also needed huge legacy RDBMS systems, buildings full of hardware and a small army of staff.
While NoSQL has changed the database industry, it has a habit of providing scale and speed at the expense of ACID. ACID is needed not just for traditional OLTP transactions but also for real time personalization/machine learning (influencing events, but only once) and managing shared finite resources (bandwidth, credit).
In this presentation we’ll show how the architecture of VoltDB allows you to have low latency ACID transactions that scale.