February 28, 2017 | Written by: Steven Teitzel
Categorized: Network Agility
Share this post:
Cognitive Service Operations – Bringing cognitive and service innovation together with network operations to drive value in cloud-based networking!
What a great opportunity this has been to take a glimpse in this series at unifying operations support systems (uOSS). We’ve touched on two of three key areas and now we get to talk to a third, cognitive service operations where we again bring together the power of cloud and cognitive to reinvent networks.
The unification of OSS comes from a service life-cycle management approach, starts with agile network DevOps but is dependent upon cognitive service operations. Service life-cycle management of a cloud-based network with NFV/SDN is dependent on executing automatically repeatable patterns of activity in the life of a service to deploy, stop, heal, scale in/out or up/down, etc. Executing these patterns occurs as closed loop continuous orchestration receives status on the health of the service, not just virtual but overall both physical and virtual elements as well as the performance of the service, who it is being performed for and what is being delivered. Conditions to execute are based on policies set as the service is developed and then ordered and deployed. These are part of the package of the service, it’s descriptions and components are deployed. But the health of the service against those policies comes from the cognitive service operations.
Cognitive service operations must be enabled by a real time topology in today’s world of physical and virtual components. What makes up the virtual instance along with the physical components must be established at time of service deployment into a topology management system. This cannot depend upon discovery later as in traditional physical systems because by”. the time the discovery occurs the service may be rescaling or changing and no longer exist. As problem determination is done, the state and components of the service must also be maintained for historical reasons to determine root causes and enable avoiding those problems in the future. Real time topology is the first step in enabling input back into maintaining the service lifecycle.
In addition, though there is much information and likely many insights on the network elements, faults and events of the network in most management systems, these need to be assembled and monitored based on the service deployed vs the elements. To do this operations and management must shift from a network focus to a service focus. This discussion could take multiple posts on why and the benefits but let’s suffice to say that a service based approach enables an evolution of network management and will help move network into the cloud world beyond the availability of components at five 9’s but of the service. This is in line with the shift of the network to the cloud and applies the world of cloud management to the network where the focus is on the availability of the service instead of the components of the network This means we must track service quality by bringing together the silos of information and insights of the network from today’s analytics into quality indicators for the service.
This move to service management in addition to network management helps to drive down key factors of cost and improve elements of customer service as well (e.g trouble tickets and MTTR). But as we work to enable the lifecycle of the cloud-based network, perhaps moving beyond reacting in the service lifecycle to things that occur, we could move to a proactive manner and scale before a problem occurs based on information of the service health vs on threshold and key indicators. Analytics tooling today will allow us to predict and with enough data scientists with network training we can set these predictors and continually update these to act upon. However, if we enable this predictive capability through the use of machine learning we can find the relation of service operations that correlate not just once but seasonally as the network and the use of the network changes unrelenting.
Here enters in the need and opportunity to apply the world of cognitive computing and cloud together to the world of network operations to get ahead of the network to ‘outthink the network’ and begin to use cognitive tooling to enable service operations. We can expand the use of cognitive tooling to augment operations as well to not only enable proactively closed loop control but also enable the operations of this changing network world. Cognitive computing (the ability to understand, learn, reason) allows us to augment operator understanding as well as find the patterns to predict upon and helps operations to focus on client and business value to rapidly resolve and improve value based operations. Cognitive service operations then comes by bringing the world of service operations needed for cloud-based networking together with cognitive computing to provide reduced operational costs but also allows us to enable effective closed loop control of the service that we track through the real time topology. Cognitive service operations allows us to bring OSS together or to unify OSS.
Join IBM at TM Forum! Live 2017 in Nice, France (May, 15-18) and see cognitive service operations in action in our private solution lab. We would be glad to discuss this more with you and talk through how we can enable this in your network environment.