When it comes managing when and where applications workloads run, IT organizations today have no shortage of choices. The hard part is determining which one makes the most sense from economic and performance perspectives, then figuring out what compliance issues need to be considered.
To make it simpler for IT organizations to navigate all those issues, Mesosphere, a provider of a data center operating system based on the Mesos kernel for abstracting underlying IT infrastructure resources, has partnered with Vapor IO, creator of the Open Data Center Runtime Environment (OpenDCRE, which provides a REST application programming interface (API) designed to provide a common method for monitoring both IT equipment and the facilities it resides in.
Vapor IO CEO Colin Crawford says the data OpenDCRE collects now can be rolled up into Mesosphere to create a currency-based schedule through which IT organizations can track their costs dynamically using a management platform optimized for container-based applications.
In that context, Crawford says every IT administrator will have the ability to become a fully functioning DevOps architect in IT environments consisting of multiple data center environments. For example, while public clouds may be less expensive for the average application workload, there are certain classes of applications in which latency issues make deploying those applications on a public cloud, which must be accessed across a wide area network, impractical. In addition, he notes, there also may be data sovereignty laws that would make deploying that application workload on a public cloud illegal.
In both cases, Mesosphere makes it possible to write rules that now incorporate both the physical attributes of a data center and the IT infrastructure installed using OpenDCRE, he says. Furthermore, via OpenDCRE actual pricing information from cloud and hosting service providers also can be factored dynamically into that decision, Crawford notes.
In general, most IT organizations are too obsessed with power usage effectiveness (PUE) ratings rather than being focused on being able to identify their true costs. While the number of kilowatts a data center is consuming is still relevant, Crawford says it’s only one data point to be considered when determining where to run an application workload.
Of course, OpenDCRE itself is a nascent effort intended to bring more standardization to cloud and hosting services. While most of the equipment in a modern data center sport REST APIs, each facility that equipment resides in tends to have its own proprietary approach for sharing information about what is occurring in the facility itself. Over time, reliance on that data tends to wind up locking IT organizations into a particular cloud or hosting services provider. OpenDCRE is designed to create a standard way to collect that data that makes it less of a headache for IT organizations to transfer workloads between multiple data centers.
The degree to which OpenDCRE will be adopted remains to be seen. But with support from Mesosphere as well as the Open Compute Project (OCP), Vapor IO is betting that, in time, cloud and hosting services platforms will be a lot more open than they really are today.