A core precept in cloud computing is that it must deliver the illusion of infinite scalability. That is, it must fulfill the expectation that any and all workloads will continuously operate at peak performance regardless of the size and number of jobs being added. In reality, however, private cloud deployments are finite with limited compute, storage, and network resources. It’s kind of like a room full of mirrors – it seems to go on forever, but stretch your hand out and you’ll hit a wall. The trick to successfully maintaining the illusion of infinite scalability is to allow workloads to extend beyond the scope of a single private cloud into other clouds and, in particular, into public clouds. Adopting this hybrid cloud approach also introduces opportunities for significant cost reductions and increased workload flexibility.
I recently came across a blog by IBM senior IT architect, Marcus Erber (http://bit.ly/1eE6Krw), which eloquently defines the scope of hybrid cloud and outlines key considerations during the planning and design phases of hybrid cloud deployment. In brief, although networking, infrastructure, and workload complexities create challenges in the orchestration of hybrid clouds, the use of standards (most notably, OpenStack) coupled with policy-based automation, such as the IBM Hybrid Cloud Solution (http://bit.ly/1k5E1T2), greatly simplify the management and provisioning of workloads, making hybrid cloud more accessible to enterprises with limited support resources. The question, then, turns from “can you adopt a hybrid cloud?” to “should you adopt a hybrid cloud?”
Here are five indications (in no particular order) that it might be time to transition your private cloud deployment into a hybrid cloud environment:
1) Constantly hitting budget limits – IT growth isn’t always linear, and unexpected cloud infrastructure expansions necessary to meet support obligations can blow an annual budget in a single month. Additionally, it is wasteful to host low priority and unrestricted workloads on costly high-performance systems. Hybrid cloud architectures allow workloads to be provisioned to the cloud environment most appropriate to the level of support required and commiserate to budget availability.
2) Can no longer expand the data center – Hard limits on data center expansion include space, power, and network availability as well as fail-over support and weight restrictions. When these limits are reached, the only options to support growth are to expand the data center (a costly proposition at best) or to move workloads elsewhere.
3) Having difficulty maintaining reliability and high-performance – Ensuring the high-availability of IT services can be very costly. It requires specialized infrastructure and knowledgeable administrators to support cluster pools, data replication, and disaster recovery. Public clouds and hosted clouds are typically architected to include these high-performance capabilities, so organizations can off-load critical loads to them without incurring additional capital and operational costs.
4) Have short-term projects, but do not want to invest in new infrastructure – Organizations can take advantage of the eminent elasticity of public clouds by provisioning services on-demand and then releasing them when they are no longer needed. In this way, private clouds do not need to be expanded to support temporary workloads only to have the new infrastructure components sit idle when the project is completed.
5) The IT infrastructure is too complex to manage effectively – In a perfect world, all private clouds would be architected from the ground up with uniform and fully integrated hardware and software elements. In reality, however, most private cloud implementations are a hodge-podge of new and legacy heterogeneous servers and storage devices bound together by virtualization, networking resources, and a variety of software components. This makes it very difficult to create standardized workflows that can be used consistently on all systems. A more pragmatic approach is to introduce a cloud management layer that allows administrative processes (supporting requirements for SLAs, security, resiliency, and capacity planning) to be standardized across all systems simultaneously. In this way, a single centralized management process is enabled that can seamlessly integrate services from multiple clouds (private or public) for a simplified management experience.
Transitioning to a hybrid cloud is by no means a scary experience. In fact, there are some very compelling reasons to jump on the hybrid cloud train today in order to proactively achieve greater efficiencies and a competitive edge. Key vendors, such as IBM, offer solutions that can simplify and optimize the adoption of hybrid clouds.
For more information on hybrid cloud and to follow the conversation, check out:
IBM “Thoughts on Cloud” blogs: http://bit.ly/1iYvzET
IBM Hybrid Cloud Solution: http://bit.ly/1k5E1T2
Hashtag: #HybridCloud #EMA
And follow my discussions on the topic at:
The Systems Café blog: http://blogs.enterprisemanagement.com/stevebrasen/