Summary Bullets:
- Two and half years ago, NTT DOCOMO embarked on a foundational NFV infrastructure project, which was commercially launched this month.
- Selecting three vendors from the original six, DOCOMO made multi-vendor interoperability and resilience key tenets of its solution, just as NFV originally envisioned.
It was almost exactly two and a half years ago, at the SDN & OpenFlow World Congress 2014 in Dusseldorf, when NTT DOCOMO committed to roll out its national vEPC infrastructure by the end of fiscal 2015, i.e. March 2016. On the ninth of this month, that indeed happened.
It was very good timing, and DOCOMO deserves to be congratulated. While there have been multiple commercial vEPC rollouts over the last year, DOCOMO set out with some distinctive prerequisites for its flagship NFV project. The company stipulated that the solution should:
- Be a multi-vendor environment, capable of extension with other VNFs;
- Use a common NFV infrastructure for scale in/out and healing;
- Reuse the existing OSS/BSS.
While these requirements may not seem overly ambitious today, at the time (2012), DOCOMO was setting a very high bar – and not just for the sake of the technology; the company’s network had suffered at the hands of the devastating earthquake of March 2011 when data loads of 50 to 60x had overwhelmed the existing network. NFV offered at least the prospect of a solution capable of withstanding a repeat occurrence. That said, even today, DOCOMO’s three prerequisites still look ambitious. Looking at the state of carrier NFV deployments as reflected at Mobile World Congress this year, number one is still very rare, number two has probably advanced the most through OPNFV’s work and number three remains a major hurdle.
Taking the first stipulation as the focus of this blog post, what’s clear is that from the outset, DOCOMO put a vast amount of work into creating a multi-vendor solution. At the height of its evaluation project, first three, then six, vendors’ vEPC-MANO/NFVI combinations were interoperability tested; no mean feat. However, when it came down to the final selection, the six (Alcatel-Lucent, Cisco, Huawei, Ericsson, Fujitsu/Nokia and NEC) were whittled down to three, with:
- Ericsson supplying the NFVO and OpenStack-based VIM;
- NEC supplying the vEPC; and
- Cisco supplying the SDN.
The only outstanding question concerns which vendor supplies the VNF manager (VNFM). Ericsson includes a VNFM in its Cloud Manager alongside its NFVO, while NEC/Netcracker’s vEPC solution also includes a VNFM. One of the hot topics in NFV today is exactly this: who is best placed to supply the VNFM, the MANO lead or the VNF vendor? Either way, in this case, significant multi-vendor integration must have taken place to get to a commercial launch, whether it was between the Ericsson NFVO and the NEC/Netcracker VNFM or between Ericsson’s VNFM and NEC’s vEPC VNF.
So, while MWC 2016 was awash with single-vendor ‘full stack’ NFV deployments, DOCOMO’s multi-vendor vEPC launch stands out as an example of an NFV deployment as it was originally envisaged. Hopefully, this new NFV infrastructure won’t be stress tested by another natural disaster, but by naturally adding more users and more vendors’ VNFs.