By Philip Merry, software engineer, SIOS Technology

Cloud computing has revolutionized how system administrators, developers, and software engineers need to consider the environment their code will run on. Code that used to be hosted on-premises on monolithic machines has been transitioned to run in cloud environments. The principles of managing such an environment have changed along with the environments in which applications are hosted. Fortunately, cloud providers do their best to help customers leverage their services to their maximum utility. Unfortunately, that help doesn't always deliver all of the requirements imposed upon those resources. With the ease of creating, deploying, and recycling virtual machines enabled by the cloud, a new wave of server administration has arrived - and for better or for worse, making servers more disposable than ever.

What is meant by a server's disposability? Literally, the term disposability means something that is intended to be thrown away. The concept is relatively simple, but it might be best to reflect on cloud best practices for background. Amazon Web Services (AWS) Well-Architected Framework  suggests the use of "Disposable Resources Instead of Fixed Servers".

Creating a server - a task that used to require a huge upfront investment of money, manpower, and planning - is now streamlined and can even be automated. IT teams can use deployment templates and cloud command line interfaces to facilitate scripting, and deploy code to manage their environment's infrastructure and hardware. They can deploy a new server quickly and far more economically by using the "infrastructure as code" model. In many cases, Cloud environments can be easier to re-deploy than they are to repair.

Unlike on-premises environments, where testing and rolling out hardware changes or software updates can be costly and time-consuming, cloud resources can be managed, swapped out, or simply re-deployed quickly and easily. Where hardware resources and the software those resources ran were once closely linked, the gap that the user sees between hardware and the application continues to grow.

One important point to keep in mind is that, while cloud compute resources are becoming disposable, companies are becoming more dependent than ever on applications and the functions those applications serve. As a result, they are placing higher and higher demands for uptime and reliability. Hardware may be disposable, but applications are anything but. Companies are more reliant than ever on applications and databases, such as SQL Server, SAP, HANA, Oracle, and others to run core business operations. As a result, IT teams are under pressure to bridge the gap between disposable servers and the high availability needed for their essential software applications.

Despite these changes, some IT principles have not changed. For example, whether on-premises on in the cloud, operating on a single server (or cloud VM) is risky, single points of failure should be still avoided wherever possible. A high availability solution is still important to ensure essential applications continue to run, regardless of the hardware (or virtual hardware) they are currently operating upon.

With a high availability software solution helping to protect applications, the state of an individual node's hardware does not correlate to a cluster's ability to maintain application uptime. The high availability software helps ensure the application keeps running, regardless of the state of the underlying server/VM. It enables IT to remove instances (temporarily or permanently), or to re-create instances without taking the application off line. Because application uptime can be retained - the application is not limited by the current system's state., but rather the readiness of the "next system" in the event of a failover/switchover scenario. This scenario presents the opportunity for IT to maintain servers in a way that is easiest to one's business - even if that means redeploying nodes. In order for business practices to grow alongside the system administration landscape, a high availability software enables administrators to treat servers as disposable while treating applications as indispensable.

##

ABOUT THE AUTHOR

Philip Merry, software engineer, SIOS Technology

Philip Merry is a software engineer at SIOS Technology, where he focuses on the company's high availability clustering and replication software products. He has a background in cloud architecture, networking, and system administration. Philip holds a BS in Computer Science from Clemson University.