Deploy your company's compute load across multiple clouds, providers and services and you'll be better-protected against complete disaster if a server fails.
That's an increasingly popular and practical notion. As a result, adoption of a multi-cloud approach, sometimes called a cloud portfolio, is growing quickly. In its 2015 State of the Cloud Report, RightScale, a provider of cloud portfolio management services, noted that as of January 2015, 82% of surveyed enterprises are now employing a multi-cloud deployment model, up from 74% just one year earlier. Within that group, a mix of public and private clouds is favored by 55%, while those opting solely for multiple private or multiple public clouds are split almost equally (14% and 13%, respectively).
As companies simultaneously move applications and data to the public cloud, keep others on premises, and integrate with software as a service providers, it's important for them to deploy services in a consistent and repeatable way. "[Fail] to work this way and IT operations will not be able to maintain control," said Bailey Caldwell, RightScale's vice president of customer success.
Consistency through automation
In its August 2015 report, a cadre of nine Forrester Research analysts states that automating is the answer to the fundamental issues of scale, speed, costs and accuracy.
Roy Ritthallervice president of marketing for IT operations management, Hewlett-Packard Enterprise
Commenting on the report in relation to cloud deployment, analyst Dave Bartoletti said, "You may have a built a workload for Amazon [Web Services] that you now want to run in [Microsoft] Azure, or replace with a database in Salesforce, or use an ERP system like SAP in the cloud. You need a consistent way to deploy this."
The problem, Bartoletti explained, is that businesses find deployment across these varied platforms difficult largely due to a lack of tools with cross-platform intelligence. "Traditionally, you'd use the tool that comes with the platform, perhaps vCenter Server for VMware vSphere environments or AWS OpsWorks to deploy on Amazon."
The tools landscape is still adapting to the reality of the multi-cloud deployment model. In his October 2015 survey of hybrid cloud management offerings, Bartoletti analyzed 36 vendors, several of which offer tools that manage multi-provider cloud platforms along with application development and delivery.
Switching between cloud environments
Consistency appears to be the keyword for existing in a multi-cloud universe. It matters because nothing stays still in the cloud for very long, including the apps and data you provide and the actual infrastructures, services and pricing of each provider.
"If you want to move applications, data and services among different providers -- and you will as part of a continuous deployment strategy -- it's important to have consistency and a level of efficiency for managing those disparate environments," said Mark Bowker, senior analyst at the Enterprise Strategy Group.
Technical reasons for periodically fine-tuning a deployment strategy include:
- Availability of new services from one provider that constitutes a competitive or operational advantage
- Difficulties with a provider
- A need to mirror deployments across multiple geographies to bolster performance
- A requirement to ensure that network communications paths avoid certain locales in order to protect data assets
- A desire to bring analytics services to where the data resides
Non-technical reasons might include changes to a favorable pricing model and the ability of one cloud provider to more fully follow an enterprise's compliance and governance requirements.
Similarly, the degree to which a cloud provider can meet regulatory requirements can lead to redeployment of applications or data from one vendor to another, said Lief Morin, president of Key Information Systems.
"When a business reaches a certain size, it has more leverage to dictate security terms to the provider; otherwise, the provider will dictate them down to the organization. It's a matter of economics and scale," he said. "In a multi-cloud environment, it gets more complicated. More providers means more risk, so it's crucial to work with them to ensure a consistent, standardized policy."
A multi-cloud deployment model should be a quasi-permanent arrangement, because nearly everything changes eventually.
"What you're seeing today is movement toward an application-configured infrastructure environment," noted Roy Ritthaller, vice president of marketing for IT operations management at Hewlett Packard Enterprise (HPE). "At the end of the day, it's not how well your cloud is organized or how shiny and new it is; it's about how well do the application and workload perform together."
While matching the application and load makes sense, the elastic nature of the hybrid cloud environment offers opportunities for continual refinement of where they are deployed, according to David Langlais, HPE's senior director of cloud and automation.
Like a swinging pendulum, a certain amount of back and forth between private and public clouds is natural, he said. "What's important is to design applications in a way that can handle changing deployment models, all the way down to managing the data and connecting to it," he explained. "Decisions that are made initially on the development side have to be handled in production for the long term. It also means understanding the cost profile and recalculating on a regular basis."
Multi-cloud management challenges
How to reach multi-cloud nirvana
Follow hybrid cloud craze with multi-cloud strategies
Public versus private cloud: Comparing cloud costs