LONDON -- OSS in the Era of SDN & NFV: Evolution vs. Revolution -- Colt Technology Services had already gone well down the path of using virtualization to streamline its service operations when NFV burst on the scene, and the company is now looking for specific values from NFV before moving forward, Simon Farrell, OSS architect, said here today.
Specifically, Colt Technology Services Group Ltd wants to see the clear ability to correlate virtual network function issues or failures with service impact, the way it does today in its physical networks.
"It's a showstopper," Farrell said. "It I can't understand the impact of an event in one VNF on the overall service, it will be very hard for us to implement NFV."
Colt came very early to virtualization, using an SDN overlay at Layer 2 to connect service resources inside its data centers as it rolled out cloud and data center hosting services, Farrell noted. The company used technology from Nicera -- now part of VMware Inc. (NYSE: VMW) -- to implement its SDN overlay connecting the edge of its data center and the individual components of the data center, including cloud resources and colocation racks, and took that live last February.
The service provider had earlier virtualized Layer 3 devices at the customer premises -- something others are looking to do via NFV today -- but this was started in 2011 and finished in 2012, just as NFV was initially being discussed. Instead of using the "white boxes" now being touted as part of NFV, Colt used Juniper Networks Inc. (NYSE: JNPR) hardware.
"NFV was not yet a gleam in anyone's eye," Farrell noted.
In addition to both of those efforts, Colt has a wide-area SDN as well, using a modular multi-vendor modular service platform to integrate Layers 2 and 3 and create an end-to-end WAN network abstraction that enables automation of dynamic Ethernet services.
Colt doesn't' want to take a step backward from its current approach, which centralizes OSS and inventory management so that service design and resource assignment can be converged and monitoring can be done across the domains and device types. That level of monitoring allows Colt to immediately correlate faults with service impacts, he added.
The company has created an abstraction layer for inventory-driven service activation so that all types of devices -- from those managed by Cyan Inc. 's Blue Planet "smart" element management system, all the way down to "dumb" proxy EMSs, Farrell says.
If an NFV orchestrator -- or any other aspect of NFV, no matter what it is called -- is able to provide a consistent view of the network topology and "talk to the level of the services" in the instance of a device or other network failure, then Farrell might be interested.
"But I think we have failed from an NFV point of view if I have to concern myself with the minutia" of correlating network faults with the affected services, he notes.
Colt is planning its own proofs of concept on this approach by the end of this year and wants it to be in production by the first half of next year.
"Don't quote me on that," Farrell said. (Sorry, Simon.)
The Colt executive also stressed the need for any move to virtualization to continue to support the five-nines service assurance of the telecom industry, and not assume that will slip to the IT industry's standards of three or four nines.
— Carol Wilson, Editor-at-Large, Light Reading