This content is part of the Essential Guide: Essential guide to cloud management
Evaluate Weigh the pros and cons of technologies, products and projects you are considering.

Overcoming OpenStack hybrid cloud integration challenges

Many IT pros still have a long wish list when it comes to building a hybrid cloud with OpenStack and public cloud providers like AWS.

Hybrid cloud is one of the top IT buzzwords for 2016. Some enterprises will opt for an OpenStack hybrid cloud, which uses the open source cloud platform for internal or private cloud operations, and one of the three major cloud providers -- Amazon Web Services, Google or Microsoft Azure -- for the public cloud segment.

However, as OpenStack is still a work in progress, the interfaces and tools available to build an OpenStack hybrid cloud aren't as complete or effective as many IT pros would like.

There's also some disconnect between OpenStack and other cloud platforms. For example, tensions exist between members of the OpenStack community when it comes to interfacing the platform with public cloud giant Amazon Web Services (AWS). Rather than making Nova -- one of the OpenStack modules -- compatible with AWS' Elastic Compute Cloud (EC2) interface, OpenStack is heading toward a native application program interface (API), and the resulting interface to EC2 lags behind AWS features. This is a big enough issue that Cloudscaling, owned by EMC, released a replacement for the existing Nova EC2 API in StackForge, a code repository, in an effort to drive compatibility.

Tensions exist between members of the OpenStack community when it comes to interfacing the platform with public cloud giant Amazon Web Services.

These challenges are likely to pop up in an open source project. And while this tension, for now, continues to make integration difficult for an OpenStack hybrid cloud, here are some key considerations and best practices for using the platform alongside a public cloud like AWS.

Configuration management in an OpenStack hybrid cloud

The lifeblood of cloud is configuration management in the form of scripts. Ideally, cloud admins could use OpenStack scripts to configure AWS and other public cloud workloads. OpenStack's Heat module aims for this, but, like Nova, it seems to be moving toward native APIs. A single tool to manage these scripts would help organizations avoid the extra work, time and inevitable manual errors that occur when copying scripts back and forth across platforms. And, from a governance point of view, synchronization between clouds is crucial.

OpenStack and AWS offer some similar services, which helps with integration efforts. For example, when selecting images, each stack has its own library of certified image solutions. Picking images that appear in both could help with integration.

Storage and performance considerations

In terms of storage, OpenStack users can deploy a broad set of alternatives to underlie its Swift and Cinder storage modules. With AWS, the choices are more limited; users can choose between Amazon Elastic Block Storage for block storage, or Simple Storage Service for object storage. While it's relatively easy to interface between OpenStack and AWS storage, it gets complicated when organizations maintain the same data in their public and private clouds.

One complication, for example, is synchronization. The interface between OpenStack and AWS is relatively slow and low bandwidth. Simple approaches such as writing data to both clouds are too slow to support many types of operations, so alternatives to protect against asynchronicity are a must.

With low WAN speeds, maintaining performance within an AWS and OpenStack hybrid cloud requires as much data duplication across both clouds as possible. This is easier to do with data that can be batch updated or that is low volatility. In addition, it's a best practice to not close down storage systems when exiting the public cloud after cloud bursting.

Data that needs to be synchronized in near real time is a different matter. Most likely, you will want to keep this data in-house, and when the public cloud accesses that data, encrypt the data stream.

Another option is to rent storage space at a telco data center, where the connections to the public cloud are faster. However, because this could slow down in-house operations, depending on the speed of connections to the telco, this model works best when the telco data center is nearby.

Tools that can help manage an OpenStack hybrid cloud

When it comes to deploying and managing an OpenStack hybrid cloud, there are tools that can help. Red Hat, for example, is extending its ManageIQ management tool to work with hybrid clouds. This should help provide a single view for configuration and setup across public and private clouds. CloudForms, also from Red Hat, aims to federate OpenStack with all the major public clouds.

The big cloud providers are also interested in the hybrid market. Microsoft now offers Azure Stack, and AWS seems willing to offer a similar "cloud-in-a-box" to its larger customers, such as the CIA. These providers also have internal tools that can help with integration, and other services and software packages will make hybridization easier over the next year or so.

There's a lot to be done. Cloud management tools lag behind the rest of the market, according to Garter. But progress is being made, and hybrid deployment -- including those related to OpenStack -- will improve immensely over the next couple of years.

Next Steps

PayPal VP explores OpenStack pros, cons

Evaluating OpenStack vs. VMware for private cloud

Discover the top OpenStack security challenges

Dig Deeper on Building and maintaining a hybrid cloud