With the introduction of SDN and NFV, CSPs are going through the biggest technology change since they rolled out carrier grade IP networks. With this change comes great opportunity for delivering services and a great challenge for OSS.
The challenge, as I see it, is that OSS must become more flexible and agile as the old rules that were built in to their processes no longer apply. Protocols are no longer fixed. Traffic routes can change in almost real time. Virtual devices can be spun up and down throughout the day. And the data center is as much a part of the service as the network is.
Can virtualization actually improve service assurance practices, and are current OSS systems ready?
I asked Phillip Doelling, Co-founder and CTO of Centina Systems….
How does a CSP respond intelligently to service assurance data from virtualized network? Put another way, with so much flexibility in virtualized networks, how do CSPs move from identifying a problem to solving it?
First, it is necessary to have a service assurance solution that can collect all aspects of virtual service performance data, from application, virtual network, and physical network and correlate this information to identify root cause as well as impacted services.
With this information, they should implement a closed-loop feedback from the service assurance solution to orchestration enabling the ability to automatically optimize services based on performance issues as well as re-mediate problems or failures to ensure service reliability.
Is service assurance a significant differentiation for new or smaller CSPs?
Yes. Service assurance of virtual services becomes dramatically more complex than traditional physical services so it can be an opportunity for new or smaller CSPs to differentiate themselves by being able to offer and ensure SLAs on virtual services better or before larger CSPs have this ability.
Are any industry standards providing CSPs with practical help in achieving service assurance of SDN? Or will SDN and NFV just further increase the ‘integration tax’ of OSS?
There are several groups including the TM Forum and OPNFV which are actively trying to define some standards in these areas. It is, however, a work in progress.
As one of the goals of SDN is to abstract network technology it has the promise of being able to reduce OSS integration efforts. However, that remains to be seen.
Do you expect SDN North Bound APIs to provide all the data needed for SDN service assurance? Or will you still require good-old NEM/NMS data and stats?
Ideally the SDN controller northbound interface provides all data needed for assurance. However, there are several needs and concerns that must be addressed for this to be the case:
- The SDN controller needs to account for all proprietary as well as generic alarm and performance data.
- The SDN controller will be a bottle-neck. This could cause issues from throughput and resiliency perspectives.
- The SDN controller must have sufficient resolution. In some cases, single-second resolution of performance data is required for high-value interfaces. It remains to be seen if controllers can poll the network and deal with large volumes of data at that frequency.
These needs and concerns are no different than EMS systems today and are reasons why NetOmnia communicates with devices directly in many cases.
Transport. Data. Applications. Datacentres. Customer devices. Is it really feasible to correlate problems across all layers and domains?
It is feasible. It requires a service assurance solution with the ability to collect all these various points of data and relate them to what makes up a service at any point in time and also rationalize this information as the service adapts and changes, but also has the flexibility and ease of use to define how data should be correlated to pinpoint where the root cause of the problem is and and what services are impacted.
Over the next couple of years, how much of Centina’s focus is specifically on SDN/NFV compared with traditional IP and transport technologies?
Centina has already proven its ability to provide service assurance of traditional IP and transport technologies and will continue to work to displace legacy service assurance systems as these technologies are not going away anytime soon.
But the trend will be toward the virtualization of many existing and new services, so Centina has and will continue to have the majority of its focus on service assurance of virtual services as they continue to evolve and be deployed.
Thanks to Phillip Doelling for taking the time to answer my questions.
SDN and NFV have been cresting the hype-curve over the last couple of years. It seems that today there is far more pragmatism around how virtualization can be deployed by CSPs to benefit themselves and their customers and, critically, how it can be managed by OSS.