Today, we announced the availability of Hyperscale instances, a new service that combines our high-end compute with 100% flash storage. Before we jump into how customers can use the service, let’s take a minute to think about the characteristics of applications and architectures deployed today.
Here are 5 key attributes we’ve noticed:
- Large volumes of data ingested in real-time. Customers and partners collaborating together; lots of machine-to-machine data as well.
- Global reach. Startups in a garage can be global in an instant.
- Mobile focus (often exclusively). Low latency and a responsive user experience is crucial.
- Highly distributed. No single points of failure or ‘master’ node controlling other components.
- Low cost of failure. Open source technologies without seven-figure contracts, minimal capital expense from hardware, if any.
Now, think about most applications running in enterprise data centers. In all 5 cases, it’s almost the exact opposite.
That’s NoSQL vs. relational databases in a nutshell. The folks at MongoDB sum it up pretty well:
If you’re a CIO, it’s simple. The more relational systems you have, the slower you can respond to the market. Relational databases will always have their place. But most of the innovation happening in the next 5 years will be on NoSQL platforms.
Of course, the public cloud is a great home for both types of systems. NoSQL apps in particular are well-suited to it.
CenturyLink Cloud hosts many different relational systems today – these are often applications that “run the business” but don’t “transform the business.” IT is increasingly moving these apps to the public cloud, freeing their time to focus on more strategic projects.
Now, with our new Hyperscale instances, enterprise developers have a service with high-end compute and 100% flash storage to make their NoSQL apps hum. Cassandra, MongoDB, and Couch all perform blazingly fast. In fact, users can expect typical performance to be 6x faster than our standard storage.
Hyperscale is a win for IT as well, since these instances are just as easy to manage as other VMs. Billing, usage, automation, and monitoring are all built-in. Full parity with Blueprints and APIs are coming soon, and additional geographies will be online soon as well.
No Pain, No Gain
But it’s not just about new distributed architectures. It’s about reducing the business reliance on relational systems.
Moving from relational to NoSQL is hard, but in many cases inevitable. In fact, this is something our engineering team has been committed to over the last 8 months. We have shifted to Couch as our primary platform, after using MS SQL for several years.
Why make this change? The same reasons many other enterprises will. More native replication, cleaner interactions with the user interface, ease of system administration, and so on.
The transition it has been liberating on so many levels. Our engineers will soon be contributing back to the open source community some important components to help enterprises do the same thing.
Watch this space for more details – and in the meantime, take Hyperscale for a spin.