Organizations of all sizes rely on cloud services to support new initiatives, stay competitive and capture greater business value. Companies often depend on multiple clouds to meet constantly evolving business goals and to avoid single-vendor lock-in and dependencies. With a multi-cloud approach, a business uses a combination of public cloud services from different providers and selects the most appropriate environment for both infrastructure and applications.
The appeal of public cloud is driven by its affordability, with a broad set of capabilities available on a pay-as-you-go basis with global coverage. Also, providers have steadily improved security and met regulatory requirements that previously prevented critical workload deployments. IT environments now regularly consist of on-premises infrastructure with an infrastructure as a service (IaaS) clone, a mix of several public, private and hybrid cloud deployments and a growing reliance on software as a service.
Organizations choose multi-cloud deployments for a range of reasons. For example, a cloned IaaS deployment ensures higher availability, more efficient failover, disaster recovery and reduced risk of crippling distributed denial-of-service attacks. Moreover, the growing data explosion increasingly requires unique approaches to handling different types of corporate information. By having more than one service and diversifying cloud deployments, businesses can more effectively match specific workload requirements to the most appropriate platform while at the same time gaining leverage for negotiating service-level agreements and contracts. However, they will also face new multi-cloud management challenges.
Multiple clouds increase complexity
Tight integration,smooth operation and consistency remain the ideal characteristics of multicloud deployments. However, managing multiple clouds generally translates to additional IT complexity. For example, controlling cloud sprawl can take up valuable IT time and resources that otherwise could be devoted to more critical operations.
Relieving IT management pressure is especially important for the growing reservoirs of data. A key challenge for IT leaders is to ensure that all corporate information in the cloud is wellmanaged and protected. IT staff often have to fulfill corporate governance requirements while still meeting strict international regulations. High levels of data awareness and management will be even more critical as the EU General Data Protection Regulation comes into play in 2018.
Multi-cloud management platforms are designed to help IT professionals unify disparate cloud environments. These platforms also enable teams to more easily execute workloads on the most compatible platform based on cost, reliability and advanced services. These are critical criteria. The faster an organization can deploy cloud resources to meet changing business goals, the more effectively it will maximize efficiency and contain costs.
Pros, cons of multi-cloud management platforms
A management platform provides organizations with new levels of order and visibility into multi-cloud environments and governance. A key goal for most businesses is to ensure that all their diverse cloud services function uniformly. As a flexible, customizable foundation, multi-cloud management platforms offer a self-service environment for requisitioning the most optimal cloud resources, from processing power to storage capabilities. Through increased visibility, IT professionals can more effectively track performance, resolve issues and accurately predict and control usage costs.
While a multi-cloud management approach provides the necessary insight, automation and control to keep cloud sprawl in check, it represents an emerging market at an early stage of maturity. That means hurdles remain. For example, large branded platforms such as IBM Cloud Brokerage — formerly cloudMatrix — VMware vRealize and Microsoft Operations Management Suite prioritize the large service providers, often excluding niche providers. These brand-based multi-cloud management platforms will function more efficiently when paired with the technologies promoted by that provider.
What should a multi-cloud strategy look like?
The challenges associated with managing multiple IaaS clouds continue to grow. If you thought keeping track of multiple SaaS bills was tough, try deciphering invoices from different IaaS providers. The problem with vendor lock-in also worsens as IaaS providers continue to expand their proprietary services and developer tools. Many IT admins are discovering they’re locked in to multiple clouds as a result of pet projects, silos or shadow IT.
Unfortunately, many initiatives that aimed to set open standards and provide cross-cloud compatibility, such as OpenStack and Eucalyptus, have lost steam. But there’s still room for hope. Amazon’s reluctant acceptance of Kubernetes solidifies a common container orchestration platform across several IaaS providers. And many function-as-a-service platforms offer unbiased access to compute resources. Organizations don’t just need multiple clouds or software delivered as a service. They need tools, standards or platforms that aim to abstract away the underlying infrastructure differences and allow multiple clouds to work like one. Whether we’ll ever see wholesale commoditization of cloud infrastructure resources is an open question, but don’t let today’s multi-cloud hype distract from this goal.
Cloud tools refocus on DevOps problems
Torsten Volk, a managing research director at Enterprise Management Associates, said the real problem is that the vendors that tried to deliver sophisticated multi-cloud management “didn’t pay attention to the DevOps guys who actually had the pain.” Over time, DevOps teams came to rely on the management capabilities of cloud providers, which also offered the ability to deliver crucial cost information.
“The key cloud DevOps problems are all related to compliance, security, cost, speed and quality,” Volk said. Third-party vendors are looking to help solve these DevOps problems by adding robust automation.
Cloudify offers an open source orchestration platform that aims to manage and automate lifecycle processes in the cloud. It is built on Topology and Orchestration Specification for Cloud Applications (TOSCA), which is a language based on OASIS. The RightScale Cloud Management platform provides organizations access to a portfolio of public, private and hybrid cloud services while maintaining enterprise control. The company’s Optima product is a cloud cost management and optimization tool. Turbonomic focuses more on workload automation for hybrid cloud environments. New Relic and AppDynamics both focus on application monitoring and management.
Other companies include the following:
- ParkMyCloud, a SaaS tool that enables users to automatically schedule on and off times for their idle cloud resources;
- FittedCloud, a platform to help match provisioning and utilization;
- GorillaStack, which helps parse AWS reports to support better utilization;
- Skeddly, which provides scheduled automation services for AWS;
- DivvyCloud, which automates optimization and compliance;
- Densify, which provides SaaS-based machine learning cloud optimization;
- OpsRamp, a cloud-based management hub; io, a cost-focused instance manager for AWS;
- Spotinst, a framework for provisioning and utilization management; and
- HashiCorp, which offers a range of tools to manage both physical and cloud-based VMs.
Multi-cloud management will remain a challenge
Regardless of who attacks the problem, there is much yet to be done in the realm of cloud management, said Dan Conde, a senior analyst at Enterprise Strategy Group. “Many people will choose a single cloud vendor for its particular strengths,” he said. But, regardless of how much management functionality individual cloud providers offer, “multicloud will be a problem.”
Primary cloud management challenges include provisioning, configuring after initial
provisioning, monitoring for cost and performance and security. “Doing things to be crosscloud may force you to be looking at or using the lowest common denominator; that would be
bad if you can’t use each cloud’s special capabilities,” Conde said.
“I think the next big area is cloud optimization, automatically ensuring that users are paying the lowest price for their resources while meeting requirements,” Rogers said. These can be categorized into two types: waste managers that continually tidy-up unused resources and broker-dealers that balance workloads across different pricing models to achieve the lowest cost. And what of the elusive single pane of glass? “It is an interesting approach if you can get people to adopt it,” Volk said.