Service providers cannot live by automation alone. Many see Software Defined Networking (SDN) and Network Function Virtualization (NFV) as a panacea for their many challenges. However, the industry has much more work to do if these technologies are to be viable. In the meantime, what are service providers to do? This three-part blog series will outline the challenges service providers are facing, discuss the requirements for developing agile wide area networks (WAN), evaluate the current SDN ecosystem, and finally provide some recommendations for how service providers should proceed.
Service providers are challenged like no other time in history. One of the main challenges is fierce competition from public cloud providers as enterprise customers change how they design and procure network services. According to Gartner, nearly half of all large enterprises will have hybrid cloud deployments by the end of 2017. Many services that were once accessible only via the company intranet are now being pushed to the cloud and made available over the WAN.
With this shift toward hybrid cloud, public cloud providers such as Amazon and Google have captured enterprise IT mindshare and budget by leveraging their brands and expertise to build large, scalable software systems using commodity technology. With rapid service activation and flexible, pay-as-you-grow pricing, they provide convenience and customization that traditional service providers find difficult to match with their legacy networks and lengthy deployment cycles.
Along with these challenges, the separation of data, application, and user presents opportunities for service providers. For example, because enterprise data is now stored in multiple places, enterprise IT organizations are seeking new transfer, back-up, and disaster recovery services. WAN performance becomes more critical, with the quality of user experience directly impacting enterprise productivity. Service providers must ensure the same experience whether these employees access applications and data over the intranet or Internet.
To take advantage of these opportunities, service providers must optimize their existing networks for multiple, disparate services. Unfortunately, this undertaking is often made more difficult with less budget and fewer engineers. Service providers must also process a higher volume and rate of requests for network resources that have to be provisioned and de-provisioned rapidly, often within seconds. This can often only be achieved with automation. As a result, most are betting on SDN and NFV technologies to enable them to build agile networks with a higher degree of workflow automation, which will also allow them to offer more services for greater ROI.
However, automation alone will not achieve their business goals. To create networks that adapt to business needs, service providers also need SDN-ready analytics for real-time orchestration and enhanced service visibility across both legacy and SDN network infrastructures. If applications change network behavior without any human intervention, real-time analytics can govern whether or not these changes are good and should be allowed. In essence, service assurance needs quality analytics to drive correct orchestration decisions. Unfortunately, SDN and NFV technologies lack the management intelligence needed to fulfill the promise of autonomous networking. Here are some use cases that depict why service providers need SDN analytics and orchestration:
Rapid service provisioning: This provides critical workflow optimization. One of the major goals of service providers is to bring down service enablement time from a week to minutes. Eliminating manual planning for tasks such as traffic engineering is critical. Intelligence such as traffic point-to-point matrices under different conditions, real-time routing telemetry and topology, and flexible network policies to generate optimization recommendations can cut down the planning effort to practically zero.
Time of the day service: Every major service provider delivers multiple services, and currently they deliver these services through over-provisioning networks for peak traffic. Alternatively, by using predictive analytics based on historical service models, service providers can optimize networks for multiple peaks over a day or week and minimize additional capital expenditures.
Hybrid cloud use cases: Bandwidth on demand and calendaring capabilities are required to efficiently use WAN resources and create new services such as cloud backup and disaster recovery needs. This requires the ability to record and baseline network routing, traffic and performance data to feed machine-learning algorithms that calculate optimum network configurations.
Virtual WAN or Virtual CPE: To meet enterprise-level service SLAs over an Internet connection, virtual WAN solutions need closed-loop network performance management, enabled by real-time telemetry and analytics.
In my next post, I will cover the evolution of WAN SDN to make these use cases a reality for service providers.