Get Beyond the Hype and Put Multi-Cloud Management into Practice
In order to maximize efficiency, multi-cloud managers need to understand the tradeoffs between public and private clouds, recognize the value of containerization, and accept the role cloud management software plays in securing the perfect service contract.
During this month’s Google Cloud Next conference, Waze Site Reliability Engineer Nir Tarcic admitted that of the more than one hundred micro-services underpinning the traffic and navigation app, “The most mission-critical ones are spread across multiple providers.” Using multiple cloud providers like Amazon and Google to “provide the best redundancy possible for users” is actually quite common: according to RightScale, 85% of enterprises currently employ a multi-cloud strategy.
Despite the prevalence of multi-cloud solutions, the management of such arrangements remains a relatively new practice. In order to take full advantage of the unique benefits offered by the multi-cloud, enterprises must operate and monitor infrastructures, platforms, and SaaS models that are intertwined in often convoluted ways. Of course, this entails overcoming a number of challenges and making several tough decisions along the way; however, if managed correctly, the resulting level of multi-cloud optimization will provide a distinct competitive advantage.
From a purely IT perspective — as opposed to a broader business perspective — the most essential aspect of a cloud computing configuration is flexibility. While all public IaaS providers offer standard compute instances, they’re all characterized by a baseline level of inefficiency. For example, when an enterprise’s primary application is CPU-intensive, but not memory- or storage-intensive, the standard instance providing the requisite CPU may well also provide excessive memory and storage — excess that the enterprise pays for whether it’s used or not.
Since they afford cloud managers comprehensive control over every functional level (hardware, operating systems, cloud platforms, and applications) private clouds are, at least in theory, an exceptional remedy for standard instance inefficiencies. Private clouds can be customized to perfectly fit any enterprise’s unique computing needs, given the right resources — however, unless an enterprise is absolutely massive, relying on an entirely private cloud tends to be cost-prohibitive. As such, running higher-value workloads domestically while migrating less essential workloads to a public cloud has become standard practice for enterprises that want to remain at least partially private cloud-dependent.
There are many ways to strategically minimize cloud costs, but sourcing services from multiple providers — that is, engaging the multi-cloud — is certainly one of the most common and effective cost-cutting methods. It’s also not unusual for enterprises to combine a private cloud with multiple public clouds. In these cases, managing the distribution of workloads across a hybridized multi-cloud environment all but demands containerization.
A special type of OS-level virtualization, containerization allows for the deployment and operation of applications without launching a unique virtual machine (VM) for each one. Without containerization, a hypervisor must instantiate a new VM running its own virtualized Guest OS for every application requesting access to computing resources. Needless to say, this requires far more capacity than a workflow wherein a containerization engine such as Docker, Rocket, or LXC services multiple applications without the mediation of a Guest OS.
Since containers do not necessitate an individualized Guest OS, they not only require less physical IT infrastructure than VMs, but they can also be moved around within the multi-cloud at will, provided the containerized applications are compatible with each cloud server’s underlying Host OS. When attempting to manage an amalgam of clouds both public and private, the efficiency and mobility of containerized workloads are truly indispensable.
The final component of multi-cloud management involves a certain amount of extrapolation and foresight. A complex multi-cloud arrangement simply cannot be cost-managed on a manual, ad hoc basis, meaning service contracts must be negotiated — and, where appropriate, amended — with an eye toward the future. That being said, astute managers understand that accurate projection is very much an exercise in evaluating an enterprise’s past and present performance levels.
Admittedly, undertaking an evaluation like this can be daunting, but the Vityl Software Suite from TeamQuest offers cloud managers access to the tools necessary to gather and organize this information. Vityl enables IT teams to collect, analyze, and monitor system performance data across on-premises servers, private clouds, and public clouds, making it the ideal software for forward-thinking multi-cloud management.
There will always be a human element involved in customizing an enterprise’s cloud computing configurations, but companies that invest in software like Vityl will discover an excellent way of aggregating the information needed to guarantee peak efficiency in multi-cloud scenarios.