AT&T’s ECOMP is Yuge
Mar 22, 2016
AT&T's ECOMP is Yuge
by Tim McElligott
AT&T recently shared its platform vision for product- and service-independent capabilities for the design, creation and lifecycle management of its Domain 2.0 (D2) environment So let’s get this out of the way from the start: AT&T’s newly proposed ECOMP architecture deserves a better name. Work this significant and game-changing has earned the right to have a moniker that doesn’t sound like it came straight outta Bellcore, especially since the new architecture is as far from the old institution as one could get. So a little creativity, whimsy perhaps, would have been nice. It may have even helped erase any niggling doubt that we are witnessing a transformed AT&T—and maybe even the last vestiges of the traditional phone industry, which Stratecast believes we are.
That said, ECOMP, or Enhanced Control, Orchestration, Management & Policy, contains everything that critics, competitive providers, and Internet upstarts-turned-conglomerates said Communications Service Providers (CSPs) could never do and represents all that naysayers said they’d never be. When the ideas are refined, the bugs worked out, the setbacks overcome and the laggards weeded out, AT&T and the industry will fulfill a lot of promises with work like this and take a whole new population of market participants and customers along for the ride. Yes, ECOMP is big stuff; it is yuge.
AT&T feels strong enough about the necessity of the architectural changes it is making and confident enough it its initial work, that the company said it is “amenable to releasing ECOMP into open source.” ECOMP might not yet be thee answer to enabling and supporting the next generation virtual network, but it is an answer. It is a well-conceived, significant answer, and it comes at a time when industry watchers were beginning to tire of asking, “Yes, Domain 2.0 (D2) is visionary and virtualization is the way of the future, but what about management and control? When are we going to get serious about security and configuration management? Performance? What about service assurance and automation?”
ECOMP shows that these issues have not been ignored. The ECOMP architecture begins to bring SDN and NFV concepts together within its Execution Framework and gives the first sign of different types of controllers and orchestrators being used within the framework. AT&T also fleshed out ETSI MANO’s service description function by adding a Design Framework, which includes metadata-driven policies for service and infrastructure descriptors and processes. The architecture also solidifies the interfaces for third party developers and OSS/BSS.
It is interesting that in the ECOMP architecture below, networking now appears as simply another part of the managed environment (grey cloud at bottom-right) on par with storage and compute resources. This is not true of course in the big picture; the network is still critical. To the contrary, this is where D2 and the AT&T Integrated Cloud (AIC) infrastructure reside. If one were to look at the D2 architecture it would look quite the opposite with ECOMP as a non-central extension off the cloud-based NFV architecture. Still it is refreshing and encouraging to see a diagram with this much detail on components that are about something other than the networking function, that puts service creation, enablement and assurance center stage.
Figure 1: ECOMP Platform Components
While Stratecast welcomes this development and applauds AT&T’s willingness to embrace the open source community, some questions are being left in limbo for the time being. Caveats were injected into conversations around OSS/BSS and automation. Regarding automation, AT&T said “a closed loop automation approach that provides dynamic capacity and consistent failure management when and where it is needed.” Stratecast agrees that it is wise not to rush into broad-scale automation, particularly with a new, software driven architecture. However, we look forward to more clarity on this front.
Regarding OSS/BSS, AT&T said ECOMP does not directly support legacy physical elements, but works with traditional OSS’s to provide a seamless customer experience across both virtual and physical elements. This, too, needs clarity over time. Key points to consider are:
- What role, if any, will remain for some traditional OSS/BSS suppliers in the new environment?
- Will there be room in the framework for independent point solutions or will existing functionality be rolled up into controllers and orchestrators?
- If the early returns from reaching out to the open source community are positive and productive, how far will open source reach into the framework?
- What of software suppliers with product built on code that is long past the sharing stage where the open source community could make any sense of it, let alone contribute to it?
Stratecast understands it is not the responsibility of AT&T or any other CSP to fret over the fate of legacy suppliers of support software. It is the software suppliers’ job to continue to innovate in order to secure their futures. Stratecast also agrees that CSPs need to get out from under their more onerous back-office business models. However, it will be good to better understand what functionality is on the table for open source and which is not. It also will be interesting to see how the gesture to the open source community is received and reciprocated.
All in all, seeing a platform with performance, reliability, security, resiliency and metadata-driven assurance at its core is a welcome approach. This is what will become known as operationalizing NFV. It is, in large part, what makes ECOMP revolutionary—even if the development and rollout over several years may feel more evolutionary.
Please add your bio info through your member profile page, or through your dashboard.