jro-grafik - Fotolia

Get started Bring yourself up to speed with our introductory content.

Tools to tackle multi-cloud management

Because no two clouds are built alike, managing multi-cloud environments can be tough. Luckily, there's an emerging class of tools that can help.

Not all clouds are created equal. While they share a common set of operations, most clouds have a quirky API or behavior that makes automated management difficult. As a result, managing multi-cloud environments becomes especially complex.

And multi-cloud complexity creates a number of issues. Most cloud vendors are eager to lock users into their service. However, vendors also realize that drastically different APIs and services capture fewer users from their rivals. Meanwhile, those users seek cloud platforms offering the lowest total cost and risk of error, as well as the best possible data and application security.

If all cloud interfaces used the same features and APIs, multi-cloud management would be easier. But, even then, cloud behavior can differ. For example, some clouds automatically provision storage when launching an instance, while others do not. There's also a lack of instance configuration standards, along with variations in software stack images and tuning -- from the hypervisor to the operating system.

What's more, security differs from cloud to cloud. Microsoft Azure prevents data persistence on instance storage, while other providers don't fully erase SSD instances. Cloud data integrity is a function of where replicas are kept, which affects geolocation, variations in local law and data replication performance.

Simplifying multi-cloud management

How do we simplify multi-cloud management problems? One answer would be a policy-based, central manger that deals with each vendor's quirks, while taking advantage of advanced features and services. This tool would provide a single, cloud-agnostic API, along with seamless flow and migration across clouds. For example, to protect key data, the tool would lock data storage in a private cloud or a colocated private storage facility. Additionally, to optimize billing, this tool would account for the costs of transitioning from one cloud to another.

The ideal multi-cloud management tool needs to monitor usage, detect bottlenecks and track billing. It would be mostly automated, but support manual intervention when needed. Tasks that do require manual operation would be made available in a policy-driven form to departmental admins, which takes pressure off cloud administrators. 

So, does this ideal multi-cloud management tool exist? Single-portal services, including RightScale Cloud Management and Dell Cloud Manager, are heading in the right direction. Some of these tools offer hands-on management, rather than fully automated controllers. But tools are evolving rapidly, as more organizations deploy hybrid clouds with multiple public cloud connections.

It's difficult to break down today's multi-cloud management tools' features. Some emphasize security, while others focus on self-service provisioning or governance. There are tools with great GUI interfaces, but they are still first-generation and, frankly, not what the job requires.

To separate the best tools from the rest, look at the products' Web descriptions. These descriptions offer information on the tools' self-service capabilities, governance, cloud bursting, backup and recovery. If the vendor describes all of these features, then it's thinking through the problem holistically -- and it's worth learning more. Additionally, the tool should work with all cloud providers -- big and small. OpenStack is also a mandatory requirement.

Once the tools are short-listed, determine which support your expected operational dynamics. You may need to re-think multi-cloud operation. Planned utilization, rigorous service mapping and fixed-priced or long-term contracts won't survive in the face of agility and leanness. Sadly, this is where almost half of cloud deployments hit a rough spot, as IT refuses to give up tried, tested and cherished management techniques. Multi-cloud and hybrid cloud require a flexible approach to control. This, in the end, is what will determine your best cloud management tool.

About the author:
Jim O'Reilly was vice president of Engineering at Germane Systems, where he created ruggedized servers and storage for the US submarine fleet. He has also held senior management positions at SGI/Rackable and Verari, was CEO at startups Scalant and CDS, headed operations at PC Brand and Metalithic and led major divisions of Memorex-Telex and NCR, where his team developed the first SCSI ASIC, now in the Smithsonian. Jim is currently a consultant focused on storage and cloud computing.

Next Steps

Balancing cloud providers with multi-cloud management

Key to cloud project success is effective management

Private cloud management tools to consider

Dig Deeper on Cloud automation and orchestration

Join the conversation


Send me notifications when other members comment.

Please create a username to comment.

What management challenges have you encountered with multi-cloud environments?
Not all clouds have the same spectrum in as much as they share a common set of operations. Some have a quirky API making automation management quite difficult. The absence of instant configuration standards coupled with software stack image and tuning variation affects the ability of the hypervisor to work properly. Security measures different with multi-cloud management as data persistence does not allow instant storage whereas others do not permanently erase SSD instances.
We have observed that the multi-cloud management platforms tend to be multi-threaded and complex. Their execution performance in production environment determines how well they would do what they are expected to do - create instances dynamically on various clouds, all the time optimizing the given objective function. Compromised performance leads to sub-optimal resource acquisition and increase in cost, latency and other factors. Any comments?

Good article and valid point, but no two workloads are alike either and another aspect of the multi-cloud environment is where the application is most suited to the platform. Emerging new tools like Krystallize Technologies Cloud QoS are designed specifically to help 'fit' the application to the right cloud service provider, or hosted environment, and to keep a close eye on the quality and consistency of the performance at the platform layer.
In the future, how a company handles their multi-cloud systems may show whether it can keep up with the fast pace of changing technology.
Hub-and-spoke solutions (one tool that attempts to adapt to all the proprietary APIs) are, at best, a stopgap solution. We need standards to reduce the variability in APIs. These standards can't (and don't have to) cover all the cases that are out there - just the major use cases and scenarios.
A "central manger that deals with each vendor's quirks...provide a single, cloud-agnostic API..." Sounds like the very concept behind the Dasein Cloud Open Source cloud abstraction library...