What specific capabilities and features should we seek in cloud orchestration software?
It's important that an orchestration platform support multiple types of virtual machines -- including kernel-based VMs, Xen, ESX and Hyper-V -- because accommodating enterprise customers' workloads often requires providing a compatible platform for them to run in. In addition, being able to create workload and application templates for reuse in separate requests can accelerate service turnup and improve efficiencies. To complement this use of templates, having an orchestration platform that incorporates DevOps tools such as Puppet and Chef as part of a Platform as a Service (PaaS) offering or one that supports an internal operational support system (OSS) can significantly accelerate deployment.
Having role-based access for the provider's teams, as well as customer personnel, is critical for ensuring that permissions for what a person is authorized to do are enforced. Audit trails are normal corollaries to these risk-based authentication policies to track their implementation. To be sure the environment will accomplish its goal, it's also critical to inspect the underlying networks running inside the data center sites as well as between them -- examining things like tenant virtual private networks and cloud data center context integrations -- to be certain the desired scaling, agility and traffic handling requirements of the service will be met.
Providers also need to align the costs of the licensing terms of the system with the expected deployments. Providers should look at the cost points for different sizes of deployments and at terms for special subsystem capabilities, such as analytics, monitoring modules and billing application programming interface (API) integrations.
In some environments, there may be additional requirements requiring extra analysis. If the goal is to orchestrate deployment of full applications from the cloud, this may include using a virtual firewall, an intrusion-detection platform, load balancing or other modules in a tenant's configuration. These may require deployment in a service-chain design specific to the application. Determining whether an orchestration platform supports creating service chains, and whether specific virtual appliances are compatible, will be important.
If a customer's application requires federation or integration with another cloud provider, it will be important to know if an orchestration platform supports federating services via pre-integrated deployment templates, such as those gateways for communicating with HP, IBM and Amazon Web Services.
Finally, the speed at which a service can be deployed or enhanced may be significantly influenced by whether the platform has truly extensible OSS and business support system APIs and a supported development environment for creating those extensions.
Do you have a question for Paul Parker-Johnson or any of our other experts? Send your questions about technical and business issues in cloud services to firstname.lastname@example.org. All submissions are treated anonymously.
This was first published in January 2014