Blogposts

Wasted cloud spend: Where is the silver lining?

In a 2019 large-scale report on cloud computing, some obvious trends could be gleaned from the statistics. The report shows that despite an increased focus on cloud cost management, only a minority of companies have implemented automated policies to address this issue, such as shutting down unused workloads or rightsizing instances.

It states that 84 percent of enterprises have a multi-cloud strategy, and that the number 1 priority for 2019 is cloud cost optimisation.

Cloud cost management and cloud governance are top challenges regardless of cloud maturity. Among enterprises, optimising cloud costs and cloud infrastructure governance are growing challenges, making cost savings the top initiative in 2019 for the third year in a row.

The challenge of managing cloud spend grows as cloud use increases.

While 64 percent of respondents cite optimising cloud spend as the most important, that number is even higher among intermediate and advanced cloud users. Other key initiatives include moving more workloads to cloud, expanding the use of containers and adopting a cloud-first strategy and implementing automated policies for governance.

Cloud users are not doing all they can to optimise costs — they underestimate the amount of wasted cloud spend.
 
Respondents estimate 27 percent waste in 2019, while measured actual waste is at 35 percent. Among Amazon Web Services (AWS) users only 47 percent use AWS Reserved Instances while Microsoft Azure users only leverage Reserved Instances 23 percent of the time.

Cloud cost optimisation best practices

The obvious way to reduce costs is to turn off unused or unattached cloud resources because they result in bills that will include charges for resources purchased but are no longer being used. Thus, start by identifying unused and completely unattached resources and removing them.

Next, consolidate computing jobs onto fewer instances. The cloud offers autoscaling, load balancing, and on-demand capabilities that allow you to scale up your computing when needed. Other cost-control techniques include using heatmaps to analyse computing demand. Automation can then be used to schedule starting and stopping of instances on weekends, for example, if no usage is expected.

Then there is right-sizing—the process of analysing computing services and modifying them to the most efficient size.

According to a Gartner report, it is difficult to size instances correctly when cloud administrators have more than 1.7 million possible combinations to choose from. This is where right-sizing tools can be used to recommend changes across instance families if necessary. 

Right-sizing not only reduces cloud costs, but also optimises the system for achieving peak performance from the resources you are paying for. 

In addition to server sizes, there are options for servers optimised for memory, database, computing, graphics, storage capacity, throughput and more. Finally, enterprises can use Spot Instances and Reserved Instances to save on AWS or Azure spend. 

Multi-cloud with a silver lining

Public cloud spend will continue to grow and double by 2023.

Cloud cost can be optimised with better management methods, but enterprises also have to consider the long-term use when deploying a multi-cloud environment.

The benefits of multi-cloud are obvious – better resilience with improved availability and lesser risks involving unplanned downtime. While much of the costs are associated within the cloud environment, enterprises have to also look at the costs of connecting to their cloud service providers (CSP). Of all the ways to connect to public clouds, one of the best ways is to establish direct connections to the CSP. 
 
Savings accrued at this level of connectivity can impact many aforementioned downstream cost controls. How?
 
An ideal direct cloud connection service provides a secure, efficient and scalable way to connect to the major CSPs. That means enterprises can scale their bandwidth (from 50mbps to 10Gbps) based on their traffic demand and choose a service term that best fit their immediate needs. 

When delivered via a software-defined networking (SDN) platform, they gain greater control and visibility over their connection to the cloud platforms.

Direct cloud connections will also result in greater cost savings in the long term. You could avoid significant data transfer fees, depending on the amount of data you transport in (egress) and out (ingress) of your cloud environment and the number of clouds you manage.

When deployed over global network fabric like Epsilon’s, enterprises can leverage hundreds of data centres with on-ramps to the public clouds from around the world as their business expands. At the same time, they can be assured of secure data transfers protected by an underlying carrier-grade private network.

By combining cloud-cost optimisation strategies with a direct cloud connection infrastructure, organisations can buck today’s worrying cloud cost trends and shake up next year’s survey results. Wouldn’t that be nice?

Author


Prasanna C

Prasanna C

Global Senior Product Manager