Resiliency is Key For Database Cloud Deployment
Deploying databases in the cloud is becoming more difficult as time goes by. Not that the technical hurdles are getting higher – indeed, technology is actually making it easier than ever – but the job of convincing the front office that this is a good way to manage precious data is getting to be a challenge.
To date, most cloud deployments have consisted of fairly low-priority data and applications. But increasingly the cloud is taking on mission-critical functions, and this trend will only accelerate as cloud-native applications and the databases that support them become more crucial to enterprise productivity and competitiveness.
A key factor in a successful cloud deployment is resiliency. The ability to withstand disruptions and to recover quickly in the event of an outage is vital the more steeped in cloud computing the enterprise becomes. Resiliency is built on top of multiple disciplines, including backup and recovery, systems failover and infrastructure redundancy, but it generally requires a well-coordinated stack consisting of substantial metadata to keep tabs on data, applications, resources and the connectivity to tie them all together.
In NuoDB’s case, resiliency is built into the automation framework and related database templates and supporting SLAs. This allows the platform to establish a self-healing “Autopilot” capability that in turn supports rapid scalability and other key functions. This is supported by an incremental enforcer retry policy and a predefined health alarm system that enable fine-grain control of the Transaction Engines and Storage Managers that populate the database environment, and thus overall system redundancy. As well, it provides ability to gauge operating conditions to circumvent problems before they impact operations.
Virtually every public cloud provider offers a high degree of resiliency, but performance can also be influenced by the way services are built and deployed. According to some, the best approach for services like AWS is to build a lean, loosely coupled environment so that individual components can scale independently. As well, you should automate the infrastructure to the highest degree possible and then conduct thorough performance testing to check for weaknesses.
Hybrid environments should also feature built-in resilience, says Forrester’s Henry Baltazar, and this will be more difficult compared to straight public cloud deployments because of the high degree of interoperability expected of internal and external resources. Top enablers for hybrid resiliency include the use of multiple availability zones and colocation sites, plus rapid cloud mobility to shift loads to available resources. But pay close attention to bandwidth charges, as even premium network services like AWS’s DirectConnect can save money over time depending on the frequency of migrations and other factors.
Indeed, the ability to shift workloads from place to place quickly and with relative ease is one of the main reasons why a multi-cloud infrastructure is more resilient than the standard data center, says GoGrid CEO John Keagy. At a minimum, a basic cloud strategy should incorporate multi-cloud backup, making sure that backup platforms and server images are consistent across all providers. More advanced deployments will incorporate tools like geographic load balancing and automated failover that enable not just high resiliency but can be leveraged to lower operating costs as well.
No system is fool-proof, of course. Fools are simply too ingenious. But the dynamic flexibility that is the hallmark of cloud computing also makes for a highly resilient, high-availability data environment that is better suited to the always-on digital culture that has permeated modern life.
Keeping data close to local resources is still a sound strategy for traditional workloads, but emerging applications are incorporating cloud functionality as a core asset. The best move right now is to ensure that the cloud you build to support them provides the highest degree of resilience that budgets allow.
Editor's Note: Learn more about how NuoDB tackles continuous availability with our new white paper, Continuous Availability with NuoDB.