Agility To Make Cloud Really Work: NPS-Demand Engineering
By Biren Mehta, Senior Marketing Manager, SP Marketing in Routing and Switching, Cisco
Search the Internet using the word “Cloud” and you will get around 1.5 billion hits and it’s not until the 2nd page or so where something related to weather comes up. This is not surprising as cloud computing is top of mind for both B2C and B2B companies. The reason for this heightened interest is because of the monetization opportunities cloud enables. For operators it’s exciting since they are challenged with exploding IP traffic and diminishing margins.
Cloud matters. The Cisco Cloud Index predicts that global cloud IP traffic will account for more than one-third (34%) of total data center traffic by 2015 and is expected to reach 1.6 zettabytes of cloud IP traffic. There are predictions about how gaming will shift to the cloud, further contributing to cloud IP traffic. As I think about emerging cloud applications and how operators will be able to monetize these trends many issues come to mind about storage and compute requirements. However, most importantly, I think about the requirements for the network, as it plays a central role in the era of cloud computing.
In the presence of significant traffic growth, today operators are squeezed. They are faced with competing challenges of differentiating services with service level agreement (SLA) guarantees while managing their costs by optimizing network capacity and minimizing operational expenses. To avoid gross over-provisioning for SLA guarantees, both proactive long-range capacity planning and reactive short-term traffic engineering processes are leveraged. However, these processes don’t natively provide a direct feedback loop between the network and applications. Network capacity and utilization remains inefficient.
To address this, Cisco has enhanced our ground-breakingNPS (Network Positioning System) technology with demand engineering capabilities for cloud applications. NPS-Demand engineering is a component of the Cisco Open Networking Environment (ONE) and one of the key innovations in delivering an elastic core solution. NPS-Demand engineering creates a direct feedback loop to enable closer coupling between the network and applications to determine the optimal location of workloads (storage and virtual machines). Demand engineering can determine where to place an application / service instance or locate content. Intelligence of network topology and traffic awareness helps identify the best location that can meet the SLA requirement and make effective use of the network capacity. Demand engineering can be leveraged in many cloud-computing applications including:
- Cloud-based xAAS offerings: which is the best data center to site an Infrastructure as a Service (IAAS) instance?
- Peer-to-peer overlays: which is the best source peer for a particular piece of content?
- Content delivery networks: which is the best cache to use for a particular piece of content?
- Web applications: which is the best application server to use for a particular request?
Demand engineering addresses the dual problem of admission control and traffic engineering. It maximizes the network utilization by optimally placing demands where there are available network resources without incurring network operational complexity, which has been deployment bottleneck traditionally. NPS-Demand engineering innovations improve network utilization economics resulting in 24% TCO savings over present mode of operation.