How to Leverage Software-Defined Networking to Increase Infrastructure Efficiency across the Enterprise
by Andrew Hillier
The concept of “software-defined” is rapidly transitioning from hype to reality. It should come as no surprise there is strong momentum of software-defined networking (SDN) adoption as organizations across the globe look to speed provisioning times across the board and eliminate any obstacle that stands in their way. In fact, Infonetics Research, now part of IHS Inc., found that 79 percent of medium and large businesses in North America are planning to have SDN in live production in the data center by 2017. However, many companies are missing a key opportunity that SDN brings to the table – one that enables them to take advantage of improved workload mobility to increase efficiency of resources beyond the network.
SDN eliminates many legacy restrictions on network provisioning and configuration, enabling private networks to be created in minutes to service applications. While most organizations are interested in SDN for the increased agility, speed and security it delivers, its full impact on where virtual machines (VMs) can and should be placed within an environment is being largely ignored. When a virtual machine can be tricked into thinking it is on the same network segment as its peers, the VM is no longer physically limited to residing on servers that are connected to a particular physical network, meaning virtual clusters and physical network configurations are no longer a limitation. This opens the door to enabling VMs to roam more freely in an environment, which presents both challenges and opportunities.
One challenge, for example, is the increased potential for VM placements to violate business policies, compliance rules and even software licensing boundaries in SDN environments. If a VM running a database server is unconstrained in where it is placed, most vendor licensing agreements will dictate the full set of hosts must be licensed to run it, grossly inflating costs. This requires a more demanding approach to VM hosting and placement decisions that consider all the utilization requirements, policies, technical requirements, and software licensing requisites. These considerations cannot just be factored into initial placement decisions, but must be leveraged to make every placement decision on an ongoing basis. To achieve this, organizations need to look at a new type of control system for workload placement: one that leverages analytics that model all of the requirements and constraints affecting a set of workloads, and uses these to determine optimal VM placements and resource allocations. This automated, policy-driven placement is essential in coping with the complexity that arises in SDN infrastructure.
There is also tremendous opportunity provided by SDN to make better use of all resource types through this new flexibility in VM placements. In all virtual environments, VM density can be safely increased by appropriately dovetailing workloads, considering their unique workload patterns and personalities. Consider the game of Tetris®. By fitting the game pieces together considering their shape and orientation, better use of the playing area can be made. Virtual infrastructure is no different. By fitting workloads together properly, density can be increased significantly and stranded or wasted capacity drastically reduced without impacting performance. When SDN is present, organizations can essentially create a bigger game of Tetris by looking beyond the cluster to optimize infrastructure to use the larger portfolio of infrastructure assets through better workload placements. This kind of optimization drives hard-dollar hardware and software savings, thus creating a strong business case for SDN adoption. Again, this kind of analysis requires advanced analytics that can dovetail workloads while factoring in all the necessary requirements and taking a forward-looking, predictive view of workload patterns to optimally determine the required placements.
While software-defined infrastructure introduces new levels of operational complexity, these challenges can be met by leveraging sophisticated analytics and purpose-built control software. SDN technologies signify an important shift in network configuration, however without the ability to codify in software all of the factors impacting VM placements in these advanced hosting environments, organizations cannot achieve their desired goals of automation, agility, efficiency and compliance.
Andrew Hillier is the CTO and co-founder of Cirba.