High Performance Database | VoltDB
page-template-default,page,page-id-13609,page-child,parent-pageid-14661,mkd-core-1.0,highrise-ver-1.0,,mkd-smooth-page-transitions,mkd-ajax,mkd-grid-1300,mkd-blog-installed,mkd-header-standard,mkd-sticky-header-on-scroll-up,mkd-default-mobile-header,mkd-sticky-up-mobile-header,mkd-dropdown-slide-from-bottom,mkd-dark-header,mkd-header-style-on-scroll,mkd-full-width-wide-menu,mkd-header-standard-in-grid-shadow-disable,mkd-search-dropdown,wpb-js-composer js-comp-ver-6.1,vc_responsive

High Performance Database

VoltDB / Product  / VoltDB Architecture – How It Works / High Performance Database

High Performance Database

The fastest in-memory operational database

VoltDB was designed to deliver real-time query responses on streaming data. A TPC-C / on-line transaction processing benchmark study conducted by Samsung & VoltDB was able to deliver 5 Million new order transactions per minute.

Read more about the Samsung and VoltDB study that analyzes in-memory database technologies here.

Here are VoltDB’s three key tenets for blazing fast performance:

The data processing landscape has changed significantly in the last decade with the rise of mobility, web scale apps, sensors, and social media; data is coming in faster than ever before. According to IBM: “90% of the data in the world today has been created in the last two years alone, at 2.5 quintillion bytes of data a day!” Legacy databases were just not built to process high velocity data.

As the dawn of 5G approaches, and the Internet of Things (IoT) becomes a reality, trillions of sensors from across the globe will transmit data constantly. To obtain instant insights from this high velocity sensor data requires a database that can process time-series data at high throughput.

Benchmarking studies show that VoltDB is the fastest in-memory relational database on the market today for high-throughput applications. VoltDB offers substantial performance and cost advantages as illustrated by the results of the TPC-C-like benchmark below.

VoltDB BenchmarkingHere’s what makes VoltDB 100x faster than traditional RDBMS at high throughput:

  • In-memory / MPP: Storing the data in-memory ensures maximum throughput. While VoltDB’s MPP architecture designed to run on low cost commodity servers enables you to scale your app seamlessly by adding servers to the existing VoltDB cluster.
  • High Availability & Durability: Unlike other databases, VoltDB never compromises on availability and durability even at high throughput. K-Safety automatically replicates partitions so that the database can withstand the loss of cluster nodes without interrupting service. VoltDB also allows you to replicate and maintain at all times two copies of your database in separate geographic locations, ensuring data safety even in the face of a regional catastrophe.
  • ACID Compliance: VoltDB’s architecture provides for no-compromise fully ACID serializable transactions even at blazing fast speeds and scale. With the ACID guarantee, app developers can focus on building and improving their app without having to worry about system failures during operations.

Learn how you can achieve maximum throughput with VoltDB here.

It is important to consider scalability as you choose a database to build or migrate your app to, that is because not all databases can scale horizontally as your app grows. VoltDB provides you with the flexibility to add more servers to your VoltDB cluster as your business grows without having to halt production. As you add servers, VoltDB automatically rebalances your data across the cluster, allowing you to scale at will on low cost commodity hardware. Here’s what enables VoltDB to scale with ease:

  • Partitioning – VoltDB automatically partitions your data and the associated processing. Multiple partitions can be hosted on a single server. Partitions enable you to not just scale up easily, but also turbo-charge performance and increase throughput.
  • Clustering – Unlike other operational databases, VoltDB does not have a master node – compute nodes architecture. All VoltDB nodes are identical, capacity and performance can be boosted by simply adding more nodes.

Learn more about VoltDB’s scalability here.

With most next generation web applications; milliseconds matter. In today’s digital economy, app consumers are demanding instant gratification now, more than ever before. Anything less than real-time is just not acceptable, and studies show that even the slightest latency can lead to shopping cart abandonment in eCommerce, downtime in industrial apps, and losses of millions by failing to detect credit card fraud in-event.

Very low latency is especially crucial in telecommunications, and most critical real-time telco apps require a “career grade” latency of under 10 microseconds. VoltDB was engineered to consistently exceed “carrier grade” latency requirements. Here’s why:

  • In-memory – Disk I/O is the primary source of latency in legacy disk based OLTP systems. Since VoltDB is a in-memory database that was architected for low latency workloads disk latency is optimized by:
    • Eliminating or minimizing excessive sequential tasks that consume disk I/O.
    • Refining Java object collections to be either short lived or very long lived.
    • Empowering admins to auto-cancel user-initiated I/O intensive tasks that take longer than a defined threshold.
  • No Latching & Locking – Research has shown that 28% of DBMS CPU time is spent on locking and latching, leading to latency. In VoltDB single-threaded processing ensures transactional consistency, while partitioning enables the handling on multiple requests at a time. The need to lock and latch is eliminated and latency is not compromised.
  • Stored procedures – In-database java based stored procedures off load the compiled data logic from the the client side and reduce latency significantly by cutting down on the number of network trips required.
  • Native Importers/Exporters – VoltDB offers fast connectors out-of-the-box for:
    • Real-time stream ingestion with either Apache Kafka or Amazon Kinesis
    • Real-time stream export to Apache Kafka HDFS (Hadoop or Spark), JDBC Databases, RabbitMQ, Elasticsearch, Amazon Kinesis Firehose, or CSV
  • Flexible configuration – Relational or Key/Value.
    At its core; VoltDB is a fully ANSI SQL compliant, relational database. However, VoltDB provides you with options to simplify your data models as necessary with the functionality of a key/value store.
  • Key/Value Store: App developers now have the flexibility to simplify the operations of their app even further by assigning values to numbers, strings, counters, JSON, XML, HTML, binaries, even images, and videos. The assigned values can then be simply accessed by a defined key.

Learn more about VoltDB’s consistent long-tail latency profile here.