Many online and enterprise applications have very spiky usage characteristics. Typical examples include e-commerce sites during holiday season, or “follow-the-sun” enterprises tracking major time zones. One of the most attractive features of cloud architectures is their ability to deliver elastic scalability – essentially capacity that can bew provisioned (and removed) on demand – to match such peaks and troughs, so that users only provision and pay for what they need and use.
Elastic scalability is critical to addressing today’s world of instant, always-on expectations, but often requires rearchitecting infrastructure to meet this “scale-out” model (e.g. storage, application servers, etc.). Unfortunately, the traditional relational database defies this model. While you do have scaling capabilities (you can upgrade your traditional SQL database to a larger machine or appliance), the database scalability is a scale up (not out) and is far from elastic: a one-time scale-up at great cost in hardware and licenses. Meanwhile, the elasticity delivered by non-relational databases such as NoSQL systems sacrifices transactional consistency and integrity.
Built from its inception as a distributed, peer-to-peer database, NuoDB was designed for both elastic database scalability and ACID compliance. By assigning peers to either transaction processing or data persistence, NuoDB effectively separates the two, enabling you to easily add transaction peers (Transaction Engines in NuoDB) to increase throughput without affecting application logic, database availability, or durability guarantees. It is equally simple to turn them off and reduce capacity – and costs – as the peak recedes. NuoDB’s tiered distributed database architecture is the foundation for genuinely elastic, scale-out performance.
NuoDB’s database for cloud-enabled global applications:
- Scales-out, not -up like “OldSQL” relational database management systems (RDBMSs)
- Scales a single, logical database to dozens of database hosts
- Runs within and across any combination of on-premises and cloud resources
- Scales-out/in elastically to instantly respond to sharp peak-load demands
- Sustains near linear scale-out on benchmark workloads