Enhancing availability, even when high availability is not required

Providing high availability (HA) protection for critical applications is a standard IT best practice. However, broader downtime protection best practices -- and options for delivering this protection -- are not as clearly defined. As a result, applications that are not considered mission critical are often left with minimal protection. Yet downtime for these non-mission critical applications can be more than a nuisance. It can impose a costly and disruptive burden on IT teams that must devote time and resources to bringing those applications back online.

Downtime can also be frustrating to end users whose work may be interrupted and put on hold. It may have downstream consequences for customer satisfaction as well.

Finding the Right Balance

In a perfect world, all IT systems would have the highest possible level of protection against downtime, but that’s not the world in which we live. With higher levels of protection come higher costs and more complexity, so we end up ensuring the highest levels of availability for only our most important applications.

For critical applications, failover clustering is the most commonly used method for ensuring HA. The application runs on one server (the primary) and another server (the secondary) is “clustered” with it using specialized software. If the clustering software detects a failure, it automatically brings the secondary server online and moves operations over to the secondary in a procedure called a failover.

More sophisticated clustering software will monitor the entire application stack -- network, storage, operating system, hardware, and application -- and automatically attempt to restart the application and even the entire server node to restore operation -- and failover to the secondary only if necessary. This process delivers 99.99 percent application uptime and can be used on premises, in the cloud, in a hybrid cloud, and even in multi-cloud configurations.

But what about enhancing the availability of those applications that are important but not critical? Simply moving these applications to a cloud infrastructure with a high availability service level agreement (SLA) may not really provide the availability protection you seek because cloud SLAs only guarantee the availability of the infrastructure, not the applications running on that infrastructure.

Put another way, cloud providers can meet their SLA obligations as long as a virtual machine is running -- even if you experience application downtime due to a software issue. You can gain enhanced availability for those important but non-critical applications, but doing so requires more than a move to the cloud.

Gaining Insight and Automation for Enhanced Availability

To keep an important application up and running, every layer of the infrastructure stack supporting it needs to be operational, compatible, and accessible. The keys to enhanced availability lie in comprehensive monitoring, ideally at every layer of the application stack. Is the underlying hardware performing properly, or has something gone awry? Is the network performing properly? Are there unexpected bottlenecks in a queue somewhere? While there are any number of monitoring tools that can provide your IT team with these kinds of insights, relying on those tools to gain that kind of visibility is only going to put a greater burden on your IT team, and that’s exactly what you want to avoid.

A better approach for increasing the availability of non-critical applications involves advanced solutions that can monitor the application environment and then proactively -- and autonomously -- respond in a manner consistent with IT best practices. When such a solution detects a condition that may result in application failure, for example, it will respond in a manner appropriate to the demands of the application. It might proactively restart the application instance, and by so doing resolve a minor problem that could, if left unaddressed, lead to poorer performance or downtime. If restarting the application fails to achieve the desired result, the software might proactively reboot the server itself. For those applications that are important but not critical, this approach provides a way to achieve enhanced availability without an investment in a more costly HA solution.

Gaining the Level of Availability your Organization Requires

Solutions providing these kinds of monitoring and automated response services are readily available. Unlike their HA counterparts, these solutions do not orchestrate a high-speed failover to another set of servers that can ensure those four nines of availability. They are running on a single server -- in the cloud or on-premises -- and monitoring and proactively managing the resources associated with a single application. If they detect a problem, they have been built to respond in a manner that is both rapid and appropriate for the application -- without needing to engage precious IT team resources to execute that response.

Indeed, a well-designed enhanced availability solution operates in the background and ensures that your important-but-not­-critical applications remain as accessible as they can be in a non-HA configuration. That can minimize the impact on your IT organization and minimize the potential disruption that your end users and customers might experience in the absence of access to these applications.

That said, if your business evolves and those non-critical applications become critical -- therefore requiring true HA protection -- it’s easy to expand an enhanced availability solution to provide full HA clustering support. You simply add one or more secondary cluster nodes to the configuration. The underlying intelligence informing an enhanced availability solution will always play a critical role -- even in an HA solution -- for it can help keep the primary production environment operating at its peak and minimize the need to resort to a server failover. Should such a failover be required, though, that’s when the HA configuration is going to ensure the highest levels of availability.

Image Credit: Michael Edwards / Dreamstime.com

Margaret Hoagland is Vice President of Global Marketing at SIOS Technology. Margaret brings to her role more than 30 years of marketing experience including 15 years of experience in the high availability space -- making her uniquely equipped to successfully develop and implement highly effective marketing solutions that differentiate and demonstrate the inherent value of SIOS products. She has a strong track record of partnering with cross-functional teams to build unity and alignment across the organization. Margaret earned a Master of Science degree in Communications Management from Simmons College and a Bachelor of Arts degree from Boston College.