Navigating the OpenStack services maze

While the range of OpenStack cloud services gives IT pros plenty of options, it can also lead to confusion -- especially in hybrid environments. Fortunately, a new tool called Project Navigator can help.

For IT staff looking to deploy cloud, OpenStack offers a lot of options. The open source cloud platform has a broad set of features, many of which can help organizations meet their cloud needs. But deciding which, out of the more than 25, OpenStack services will best suit your cloud environment can be tough.

Adding to the confusion are the hundreds of vendors that support OpenStack, offering a variety of distributions and additional tools and features. Meanwhile, hybrid cloud adoption makes all of this more complex and exciting. In addition to the challenges of building an OpenStack-based private cloud, hybrid cloud introduces new hurdles, such as bridging servers, networks and storage across clouds environments.

Matching the right OpenStack services to your cloud needs

Let's tackle the simpler task first -- building a private cloud with OpenStack. The first issue admins face is choosing from the wide range of OpenStack cloud services available. Luckily, OpenStack has released a tool, called Project Navigator, that helps organizations determine which OpenStack services they need to build a cloud for a particular use case, such as Web serving.

The Navigator pulls together a dashboard of status information on each OpenStack project module. It also shows project maturity, which is essential, since new modules are added regularly. The tool breaks out OpenStack services into six core modules that all OpenStack clouds should use, and then optional services for specific cloud use cases.

This categorization is likely to change over time, as more of these "optional" services reach full maturity. At that point, more OpenStack services -- such as the Horizon Dashboard, Heat Orchestration, Magnum Containers, Congress Governance and Barbican Key Management -- may enter the core service pack.

By providing details on OpenStack services, Navigator helps admins make decisions about specific workloads or use cases -- but OpenStack configuration help doesn't stop there. There are also sample OpenStack configurations aimed at making sandboxes easy to implement. These sample configurations are based on the experience of major OpenStack contributors, such as CERN.

Using OpenStack services for hybrid cloud

For reasons such as cloud bursting or backup, most organizations want to use both private and public cloud. So, while Navigator is an excellent way to kick-start a private cloud implementation, it might not be enough for hybrid deployments.

When you cut through all the hype, it's still more difficult to build a hybrid cloud than it is to deploy a private or public cloud separately. This is because hybrid cloud requires networking structures that cut across the boundary between public and private, as well as storage systems that are positioned to get performance from either.

When you cut through all the hype, it's still more difficult to build a hybrid cloud than it is to deploy a private cloud or public cloud separately.

These concepts are still in their early days, and pose some fundamental challenges. For instance, slow wide area network speeds make it difficult for public cloud instances to read data in private clouds. What's more, some larger enterprises split their private cloud deployment into different geographic zones, which have to be federated together with public cloud zones.

For an OpenStack hybrid cloud, it's essential to bridge VLANs between cloud environments, while maintaining security and authentication. There are some technologies, such as OpenContrail, that can help.

Despite the challenges, there are still options for creating a hybrid cloud with OpenStack services. For instance, some organizations use OpenStack and Amazon Web Services (AWS), as OpenStack Heat scripts are very similar to AWS scripts.

The future of hybrid cloud will evolve with software-defined infrastructure (SDI). SDI plays to the advanced orchestration that will bind federated cloud segments together. With the control plane services abstracted in SDI, it is possible to build higher-level APIs to allow seamless cloud-to-cloud operation.

We are still some time away from having a fully automated orchestration process that handles apps and data, as well as platforms. This is one of the most interesting parts of the cloud evolution and it's getting plenty of focus from developers. As a result, using OpenStack services in a hybrid cloud environment should become less painful over time.

Next Steps

How to find OpenStack development support

Comparing OpenStack vs. VMware for private cloud

PayPal VP shares pros, cons of OpenStack deployment

Dig Deeper on Open source cloud computing