Where's NaaS?

Networking as a service can be offered in exactly the same way as the other "as a service" offerings, like software as a service (SaaS) and infrastructure as a service (IaaS). However, the current industry situation in SA is that NaaS implementation is being held back by the requirement for long-term connectivity contracts to enable it.

It's clear this situation needs to change if SA is to see true NaaS roll-out.

NaaS services are very different to the networks seen today, which has some good and bad implications. I'll use the South African market to describe how a NaaS offering would look and what's essential in such a network environment.

1. The NaaS provider will need a traditional countrywide core network, which is generally MPLS. This is essential, as the NaaS provider still provides every component that a current MPLS provider does, with the exception of the last mile.

2. Important, but not essential, to this design, is the ability to integrate this network into cloud application networks such as local cloud or global cloud. Almost every company or organisation is considering, rolling out or already working in the cloud.

3. The NaaS provider needs an overlay software capability that will allow it to:
a. Launch a VPN tunnel from inside the clients' branches to the NaaS provider edge. This can be done using IPsec, or in some cases, proprietary VPN tunnels as used in ViBE;
b. Control application delivery from inside the network, over the Internet into the core, with application steering or QOS;
c. Use intelligent algorithms that deal with packet loss and latency that can arise in the Internet last mile, such as forward error correction; and
d. Route all Internet VPN tunnels into a single customer VRF to create a countrywide or global VPN network.

The benefits of NaaS are not necessarily as clear as would be preferred, largely due to the use of the Internet as the last mile connection. Aside from this, the benefits are numerous, including:

1. Providers can roll out a new network in days instead of months;
2. Sites can be added to a network in days instead of months;
3. Network costs are generally lower than MPLS - in some cases, they can be up to 10 times less expensive;
4. As long as the provider edge node is close to the branch office, quality over this network should be good, as long distance data is travelling over a solid core;
5. Adding cloud services is simple and easy; and
6. There are lots of new capabilities in these networks not available in MPLS, such as dynamic link selection, packet loss mitigation, application awareness, TCP window improvements, network orchestration, and central deployment and configuration management.

The best alternative is to separate the last mile fixed links from the service provider network. In other words, to buy the last mile links from someone other than the NaaS provider. This has become possible due to data centres like Teraco, where every provider is represented - the last mile provider no longer has to be the NaaS provider. However, this isn't as effective as the Internet, in terms of flexibility and cost-effectiveness.

Contracts are the largest obstacle to moving to a NaaS model.

So, why don't large service providers offer NaaS?

Several easy-to-understand reasons will keep NaaS out of reach of many companies if they expect to get it from their service providers. These include:

1. Contracts are the largest obstacle to moving to a NaaS model. A NaaS service model means the customer can leave the NaaS provider in a matter of weeks, and vice versa;
a. Service providers want long-term contracts to protect them in what is a relatively poor service industry.
b. Customers want contracts for the same reason. Large enterprises use contracts and binding SLAs to keep service providers in check, and often recoup much of their spend through service penalties.

2. Large providers are heavily invested in MPLS and want to protect the associated revenue, which remains their largest source of income;
3. It's not easy to define SLAs when the access circuit is Internet-based, and customers want end-to-end SLAs to ensure network performance; and
4. Customers with large branch networks don't want to separate the last mile from the network, or don't want to be responsible for the access network (Internet) at each branch. They want to outsource this, usually to a single company.

NaaS is a clear and maturing industry, with companies redefining the way networks are delivered. Over the next few years there is a real possibility that the gaps between IaaS and NaaS will reduce or be removed, opening up the opportunity to transform networking, as IaaS and SaaS did with hosting and application adoption, respectively.