UNS Spotlight on Dynamic Workload Scaling with ACE and OTV
In an earlier Unified Network Services (UNS) blog update, we highlighted application performance monitoring with NAM and the important role it played in optimizing WAN acceleration deployments with WAAS. Today we are going to focus on another aspect of application performance and quality of service that we call Dynamic Workload Scaling (DWS) or Cloud Bursting. The basic scenario is that one of our mission critical data center applications experiences intermittent (and potentially unexpected) peak loads that can overwhelm the server resources that are typically required for the quality of service (QoS) that we require. An example might be an online flower store that gets two orders of magnitude more traffic in the weeks of Valentine’s Day and Mother’s Day than the other 50 weeks of the year (although, in this case the expanded capacity is largely predictable).
Fortunately, this is a virtual application and we have the ability to create new virtual machines to handle additional capacity. Many organizations are now looking to outsource these peak loads to cloud service providers rather than owning all the resources that might only be used 10% of the time. In many cases, that would be a lot of cap-ex to waste. Or maybe there’s an in-house remote data center with additional capacity that be leveraged, i.e., a private cloud. Either way, public or private cloud capacity expansion, provisioning and load balancing presents some challenges, not the least of which is figuring out how to optimize the use of these cloud resources so as to minimize costs, and effectively distribute loads across all available resources, including multiple cloud locations.
One of the primary challenges is expanding the layer 2 domain of your data center network across multiple cloud sites, since this is required for efficient movement and expansion of virtual workloads (applications), as well as for load balancing. Cisco created Overlay Transport Virtualization (OTV) on the Nexus 7000 as an optimal way to expand layer 2 domains between sites that would normally require layer 3 routed connectivity. OTV allows remote cloud sites and resources to operate as efficiently as if the resources were local to one data center.
We have created a slick video to visualize how our ACE Application Control Engines can take advantage of OTV to expand application capacity to these remote cloud sites automatically, based on quality of service parameters set in ACE. At certain capacity thresholds, loads will only then be sent to the cloud, and ACE will continue to balance workloads to the expanded sites until the baseline resources can once again handle the prevailing requests. This ensures that cloud resources are optimally utilized, limiting costs as much as possible. Dynamic Workload Scaling or Cloud Bursting is a great example of how UNS applications (layer 4-7 services) like ACE offer a superior solution by being well-integrated with the network fabric (layer 2/3 services) and the rest of the Cisco Data Center Business Advantage architecture. Sound interesting? We’d love to hear your comments on the video and what you think of DWS and how it could facilitate your cloud expansion plans.