Skip to main content

How to get cloud optimization right

(Image credit: Image Credit: Everything Possible / Shutterstock)

Deployment of any new strategy or tool has never been more under the spotlight as it is now. As businesses have experienced a (hopefully) once in a lifetime trifecta of challenges, including Brexit, Covid-19 and potential global recession, budgets are under intense scrutiny. However, for those that can, technical innovation and adoption of decentralized public cloud services will be a key differentiator in adapting to the ‘new normal’.

For those considering a cloud deployment, effective cost management will be critical to success. Initially, this will mean making the right choices on a per-workload basis in terms of public/private cloud placement and migration strategy. But to continue maximizing the benefits of cloud adoption, early action in three key activity areas will be required to enable effective cost optimization as your cloud footprint grows: monitoring, metadata and, most critically, accountability. These areas will impact not just your infrastructure, but the people and processes that deliver your IT and financial management capabilities.

While it might feel that a significant amount of resource is required before the deployment has happened, the activity will establish the foundations for the cloud platform and approach that you take. It will also ensure that an organization holistically considers its needs – to suit it both now and in the future. Not taking these elements into account during the journey to the cloud can not only impact the initial migration, but subsequent experience with ongoing operation and adoption of the cloud – both in terms of ability to innovate and management of ongoing costs.

Monitoring

It should go without saying that an effective monitoring strategy is critical to a successful operational strategy for any platform. Metadata monitoring underpins capacity management, performance management and incident/problem management. However, effective monitoring is also key to both a cost-efficient migration strategy, and ongoing cost management within the cloud.

Effective gathering of infrastructure and application performance metrics provides the key to understanding your existing footprint and its performance requirements, enabling effective design of the target cloud footprint. It is also a useful indicator for the adoption of cloud-native PaaS offerings to offload potential performance bottlenecks or management headaches to your public cloud provider of choice.

Performance metrics can also identity unexpected potential cost savings, allowing you do discover, for instance, that a virtual machine with almost no CPU or network traffic might be a target for deprecation as part of a migration exercise. This is a question raised by your performance metrics that will be resolved when you come to metadata management.

Your monitoring strategy will also continue to play a key role in cost management post-migration. In addition to its existing roles in performance, capacity and incident /problem management, your performance metadata will aid in identification and resolution of a cloud-specific management challenge: overprovisioning.

Adoption of agile methodologies, fail-fast and rapid prototyping in the cloud can yield real business benefits, but the decentralization of infrastructure management required to realize these benefits can yield real challenges. The pressure to deploy rapidly can result in a tendency to overprovision resource, as it enables a ‘deploy first, scale later’ strategy for rapid go-live, and enables upscaling of resource to handle potential post-update performance bottlenecks.

In both of these scenarios the intention to right-size once the resource requirement has been identified, or the performance bottleneck identified or resolved, are often quickly superseded by rapidly evolving business priorities. For example, overprovisioned resources often remain overprovisioned and incur additional costs. In these instances, effective performance monitoring, coupled with appropriate accountability, will be key to ensuring that these overspends remain visible and appropriately prioritized.

Managing metadata

Establishing appropriate metadata on infrastructure and applications is another key area which will affect both migration planning and subsequent cost management in a growing cloud infrastructure.

If your organization has a robust and well-annotated configuration management database covering all infrastructure ownership and dependencies, congratulations, your job in planning a migration just got an awful lot easier! However, configuration management in most organizations will be at differing levels of maturity, and more often than not, most organizations will be running workloads which have grown organically, are poorly documented, or were put in by a contractor who left a while ago and haven’t been touched since.

If your workload metadata is not all it could be, migration planning is the time to resolve this and establish:

  • Who owns the workloads from a business perspective?
  • Who owns the workloads from a technical perspective?
  • What project, application or service do your workloads support?
  • What are the upstream and downstream dependencies for your workloads?

Answering these questions is key to your migration strategy. They will inform who needs to be consulted or informed on migration events, who will be best placed to test migrated workloads, and if any upstream or downstream services may need to be co-migrated. Coupled with performance metrics, this information will also inform whether a workload needs to be migrated, replaced or even retained within the business.

Post-migration, maintaining metadata and assigning it to new resources becomes even more critical. Unless key ownership and application metadata is enforced in your new cloud environment, costs can rapidly become unattributable and difficult to address. Only when coupled with appropriate attribution metadata can performance metrics be effectively assessed to identify and implement cost-saving measures.

Initial cloud migration is the ideal time to be adopting a metadata enforcement strategy, before your cloud footprint begins to grow. As with oversizing of cloud resources, appropriate metadata tagging can all too often be a casualty of tight deployment deadlines or changing business priorities, and when it comes to ensuring attributability for cloud workloads an ounce of prevention can be worth a pound of cure.

Establishing accountability

Having initially established and enforced maintenance of key ownership and functional metadata about your cloud workload, it’s time to address the thorny subject of cost accountability in the cloud.

Enabling innovation is typically a key driver of cloud adoption, with the ability to innovate, iterate and fail-fast among most tangible and transformative benefits realized by successful cloud adopters. To effectively embrace these changes, an adjustment in thinking about how infrastructure costs are handled is required.

The traditional model of IT procurement (where resources are centrally scoped, purchased, owned and managed), does not address the new reality where developers from different business units are empowered to provision cloud resources to enable innovation and rapid deployment. A frequent scenario in cloud early adopters is a model in which business units have the ability to develop and deploy their own cloud resources, but cloud costs are still centrally billed and managed.

In this scenario the central team retains the responsibility of ‘cost optimization’ and is tasked with identifying potential cost saving measures from performance metrics. They must then identify and chase down the resource owners to request validation and implementation of the proposed measures. They invariably find themselves at odds with a team tasked solely with delivering a performant service and no incentive to make the requested changes.

The result is all too predictable: inaction on cost optimization measures and tension between central IT teams, finance and the lines of business owning resources, lead to spiraling cloud costs and disillusionment with the cloud.

The answer is obvious from the outside: the business units requiring the ability to develop, deploy and manage their own cloud services should be accountable for the costs incurred in their development, deployment and use. However, this requires a business transformation of systems, process and people that impacts not just the IT function but also lines of business requiring the capability and the finance teams involved in recharging. Unsurprisingly corporate inertia often results in this necessary transformation being either deferred or ignored entirely.

The common result is unchecked cost increases resulting in either a retrenchment from cloud or a recentralization of infrastructure management. Either outcome results in a whole or partial loss of the very innovation and agility benefits that cloud adoption was introduced to deliver.

The simple conclusion? Grasping the nettle of accountability for public cloud costs early in the cloud adoption process is critical to enabling both realization of the benefits of cloud and effective management of cloud spend.

What next?

A combination of monitoring, metadata, and accountability are critical to ensuring that your cloud adoption or migration is a success, both in terms of project delivery, and ongoing cost management in the post-migration to the ‘new normal’. Using an experienced partner to support your organization’s journey to the cloud can be hugely beneficial, not just in augmenting your skills to enable the design, planning and successful migration to the cloud, but in supporting you in delivering the people, process and supporting service transformations required to realize the benefits of your ongoing cloud adoption.

Dr Peter Cridland, Head of Cloud Enablement, Brightsolid