Sasa Nijemcevic (pictured), vice president and general manager for the Network and Service Management Business Unit at Nokia Networks, explains how integrated assurance can help new SDN services run at peak efficiency and meet SLAs
SDN Insights: Automation and optimization have been key selling points for carrier SDN - why is that not enough for delivering SDN services?
Sasa Nijemcevic: As SDN projects move from limited live trials to broad deployment and commercialization, providers are discovering that carrier SDN platforms must also support integrated assurance capabilities. Assurance has always been a key part of service fulfillment, but in a world where the dynamic network consumption patterns of cloud applications makes network demand and traffic patterns less and less predictable, the ability to ensure your network and services are doing exactly what they are supposed to becomes more important than ever.
How can integrated assurance capabilities help?
By integrating assurance with carrier SDN, providers can ensure their new SDN services are operational, requested services SLAs are always met, and the underlying network infrastructure is always running at peak efficiency. For instance, service fulfillment can be validated with integrated OAM tests at the end of the SDN service provisioning process. Insightful analytics such as KPIs and correlations can be used to quickly identify possible service impacts from changing network conditions, and allow operators to make whatever changes are necessary to ensure SLAs are maintained. Assurance data also provides much of the visibility and intelligence required to make smart services placement decisions in an unpredictable environment where network resources are consumed and released on demand. What’s not as obvious is that while assurance can help carrier SDN advance to the next level, the converse also holds true. When mated with SDN automation, assurance processes can become just as dynamic and efficient as the SDN networks they support.
What has to happen to make assurance processes just as dynamic as the SDN networks they support?
This is where taking a holistic approach to carrier SDN is a clear advantage. The analytics/KPIs/correlations that drive assurance GUIs and visualizations today must be integrated with the SDN controller so that they can drive dynamic changes to the network. This is often referred to as closed-loop, or dynamic assurance. Assurance policies can be used to trigger actions at the services layer to ensure critical SLAs are met, and at the network layer to ensure serious network congestion is avoided and optimal use is made of network assets. A comprehensive approach is required to provide optimal flexibility so that flows can be redirected, new IP/optical paths established, and existing IP/optical paths can be resized dynamically - all driven by KPIs/analysis/correlations from both IP and optical layers, and from both physical and virtual domains.
How should operators be thinking about implementing this?
It’s really about figuring out your pain points and zeroing in on these rather than trying to boil the ocean. This has been the strategy at our most successful customers. For instance, one is focusing on delivering dynamic VPN services and is leveraging dynamic service assurance to check bandwidth utilization and resize paths as necessary at the IP/optical layer (i.e., by adding and Ethernet link to a LAG) as part of their SLA offer. Another use case is leveraging dynamic network assurance by checking congestion on multiple links/paths to and from datacenters and dynamically remapping flows to ensure that link efficiency and customer satisfaction remains high - all without having to make constant changes to their routing tables.
What are the key challenges in getting this right?
From an implementation perspective, getting dynamic assurance right sometimes requires tying into a broader assurance model, and slowly building trust in the policies that drive automated assurance processes. From a broader model perspective, it means being able to abstract the KPIs/analytics/correlations that come from the multiple domains the carrier SDN platform controls, and feeding that up to the IT/OSS layer which combines this data with analytics outside the carrier SDN controller’s domain. The IT/OSS layer can then decide when to trigger abstracted assurance polices at the SDN layer, based on its analysis of the broader set of data.
This article was first appeared on Telecom Asia SDN Insights October 2016 Edition