Cloud Control: How Fast Is The Migration To The Public Cloud?

Survey Says…

Cloud Cruiser, a developer of cloud monitoring solutions, set out to capture some hard data on the pace of migration to the public cloud and the ability of users to track and control cloud consumption and cost allocations.

To that end, the San Jose-based vendor sponsored a survey that Dimensional Research conducted among attendees of Amazon Web Service's Global Summit in Chicago.

Researchers handed out questionnaires at the event in April to new and existing AWS customers, asking them to share information about their organizations' usage plans, methods, and struggles in running and operating cloud deployments.

189 IT professionals completed the survey. Here are some of the interesting findings.

AWS And What Else?

The Cloud Cruiser survey asked AWS Summit attendees what cloud environments, if any, they used in addition to Amazon's.

More than half -- 54 percent -- said AWS was their sole cloud provider.

But among those operating multi-cloud environments, Microsoft was the most-popular choice, with 27 percent of the 189 respondents revealing that in addition to AWS, they were running workloads on Azure.

Another 22 percent of respondents were supplementing Amazon with some private cloud usage, while 12 percent had deployed Google Cloud Platform. And five percent used a variety of other providers.

Eight percent of the AWS Summit attendees weren't using any public cloud provider service, including AWS.

Mohan, of Mindtree, told CRN that the increasing use of multiple cloud solution providers is placing on IT departments "increased scrutiny to measure, justify, and govern cloud usage."

Ramping Usage

The AWS Summit attendees were asked how much they expected to increase cloud usage over the coming year.

A 40 percent plurality said they expected to ramp their public cloud use between 25 and 50 percent within the next 12 months, while nearly as many -- 38 percent -- had a more conservative target in mind of between 10 and 15 percent.

In all, 95 percent of respondents estimated at least some increase in public cloud use, with 17 percent expecting an increase of less than 10 percent. Only 5 percent said they were keeping their cloud deployments steady.

The survey "confirms what Mindtree has experienced," said Mohan. "Cloud is standard and usage is expanding due to development/testing and big data/analytics. The latter in particular is greatly accelerating cloud usage."

Big Data Driver

The public cloud, when first introduced by AWS, found a vanguard of adopters looking for a flexible environment in which to conduct software development and testing. Those types of non-production workloads are still the most-common use cases among Summit attendees, deployed by 60 percent of respondents polled.

But the cloud has also ushered in a revolution in big data and analytics, and those workloads are increasingly common -- 31 percent of respondents were running big data applications in their Amazon cloud environments.

"Participants indicate that public cloud momentum is now being fueled by big data and analytics," the Cloud Cruiser report found.

Another 30 percent were using their public cloud as a sandbox for experimenting with new applications. One-quarter of respondents were running customer-facing applications, 19 percent were using AWS for backup and recovery, 14 percent were running business applications, and 2 percent were conducting eCommerce on the platform.

Management On Their Mind

More than half of the survey's 189 respondents said cost allocation and reporting were important concerns when managing their public cloud resources. Both performance management and optimizing resources were considered important by just under half of those polled.

Troubleshooting spikes in usage or costs, forecasting and budgeting, and managing reserved instances were cited as important concerns by between 21 and 23 percent of respondents, with 18 percent also citing monitoring service pricing changes.

Management Shortcomings

The Cloud Cruiser questionnaires asked respondents about the tools they used to proactively manage their public cloud spending and usage.

A 31 percent plurality responded that they didn't track, manage and allocate costs and consumption at all. Another 17 percent did so with a spreadsheet or did it manually.

Mohan, of Mindtree, told CRN that with 17 percent of companies using nothing more advanced than spreadsheets, and 31 percent of companies ignoring spending and usage altogether, compounded with increasing cloud use, "almost half of IT will struggle to manage cloud cost and cloud governance."

A quarter of the respondents said they used a custom system built in-house to manage their cloud systems. Another 20 percent used the tools offered by their cloud provider, and 16 percent purchased solutions from third parties.

"The top challenge with the continuously growing public cloud resource is the ability to manage allocation usage and costs," the report stated.

Deciphering The Bill

Cloud billing statements aren't always easy to make sense of and 82 percent of the respondents, to various degrees, said they found it difficult to reconcile the invoices they received with their financials.

Almost 80 percent categorized the process of reviewing and approving their cloud bills as "just okay" or challenging. And 3 percent said the word "horrible" aptly described the process.

Only 18 percent found reviewing their cloud service bills to be easy.

"IT and finance continue to have difficulty working together to ascertain and allocate public cloud usage, and IT continues to struggle with technologies that will gather and track public cloud usage information," the report stated.

Tagging Troubles

Tagging is a technology necessary to attribute cloud service usage, but more than one-third of respondents polled at the AWS Summit said they weren't using tags at all.

Another third of respondents that were using tags said they struggled with the problem of a lack of policy enforcement. And 21 percent said they didn't know what type of tagging strategy to implement.

Another 16 percent had the problem of their tags being inconsistent and error-prone, and 7 percent of respondents struggled in retroactively correcting bad tags.