Apr 26, 2017 Going for Broke(r): Putting More Cloud in Hybrid
The hybrid service delivery environment wants to be a true hybrid cloud. One thing it needs in order to get there is a good cloud service broker.
What Is Hybrid Cloud?
Not what we are doing now. What we are doing now is, hybrid service delivery: we have solutions coming from a bunch of different places along the sourcing spectrum–our own data centers and colocation spaces, managed hosting and managed application services, private external clouds, and public clouds for IaaS, PaaS, and SaaS. But, we don’t integrate consistently and we don’t have holistic management or monitoring and we don’t generally have easy workload mobility across sourcing options. When we want to have an application served out of our own data center we deploy it one way, when we want it running out of a hosted data center we deploy it another way, and when we want to run it out of Amazon or Microsoft or Google, we have a different deployment approach in each. We may even have different people doing the work in each case.
Hybrid cloud demands a real private cloud in the enterprise data center, first, and process and data integration among environments. It also needs holistic monitoring and management, and orchestration across environments including smooth and automatic mobility of workloads, with each workload placed in the infrastructure that can best serve the enterprise’s needs.
Cloud Service Broker: Right Place, Right Time
This last aspect of a true hybrid cloud is the province of the cloud service broker (CSB). CSB’s have the job of providing a degree of abstraction across environments in order to ensure the movability of workloads. They vaccinate against cloud lock-in by presenting a palette of low-level entities (like VMs) and services (like security grouping) that can be instantiated in each environment and allowing cloud solution designers to build new services out of them.
Cisco CloudCenter (nee CliQr), RightScale, and Scalr work along these lines. CSBs are not widely deployed yet, but as more organizations shift larger and critical production workloads into public-cloud IaaS, more will also deicde to implement CSBs.
Ideally, a CSB will go a step or two further, of course: not just allow you to deploy into different environments but to model the use of those environments in advance, for planning purposes. That way, you could see in advance what you could expect from placing a workload in one versus another. Would you get better performance at a higher cost for job A in Google? Lower costs but much higher latency in Amazon? The more ways the tool can model the realities of using different environments, the more informed IT’s decisions can be about where and how to meet a specific need. Cost is a critical consideration, but others, such as performance and compliance, can be equally important to control and to factor into capacity planning.
As enterprises work towards the tipping point for hybridization, when on average half or more of their workload is running in public cloud contexts (sometime in 2019, we think), the capabilities of a CSB will become not just desirable but crucial. Whether as SaaS or in-house installation or itself a hybrid, the CSB will be central to how IT gets done.