In this third piece in my series on a recent Colt Webinar on uCPE we look to the future prospects of uCPE and what this technology means for service providers (SPs). You can find part 1 which is a more general over, here, and part 2 which looks more at the deployment case study presented in the webinar, here.
Virtual VNFs from the app store
On the webinar, Colt discussed all of the virtual VNF’s that can be deployed to its uCPE and its plans to expand that third-party library of software applications (e.g. firewalls, optimisation, routing, analytics, etc).
Currently this needs to be configured and rolled out by Colt, but they are looking towards a future where the uCPE device operates like a smartphone app store. This means the customer will be able to see the operating resources they have left, memory/processing etc, and download any applications they want on demand.
I think this is a pretty optimistic goal but not unreasonable in the long term, we have seen the capabilities of generalised hardware grow at every turn, with more compute, memory and flexibility being added at every turn. We have seen through the growth of SDN and SD WAN, that the virtualisation of network functions is not just possible, it is highly practical.
However, there is a big challenge with this optimistic view in the short-term. In the last article we discussed the cost of the technology and how buying the capacity to have this “App store” like functionality is expensive and may go unused. After all, if we hijack Colt’s analogy how many people buy that 256GB iPhone and only ever use 50% of the memory capacity. Did those people need to spend 20% extra to get that extra capacity? In the consumer world that’s only a loss of £150, in the business world we start adding zeros. Yet this analogy continues to prove useful, because 256GB storage on a phone 10 years ago was unnecessary and unachievable. The same cost vs capacity challenge will eventually shift for technologies like uCPE and they will become more widely available.
In the meantime, SPs should focus on the larger enterprises who have specific uses cases where uCPE enables new business strategies and critical operations and because of that the money to fund the systems will be made available. I would also expect to see some success in selling packaged uCPE products that offer limited flexibility, but come with pre-set functionality that replaces outdated hardware in a cost-efficient bundle. These packages will play well in the SMB space where legacy hardware is cycling out, and companies want to replace 3-4 specific services and don’t expect those service needs to change drastically over the mid-long term. These companies will benefit by having one box instead of four, reducing expensive contracts, utilities overheads and admin.
However, there are also risks to this approach. SPs selling packages must now be able to identify demand for and support a bigger variety of solutions, understand how they fit into the network and have the support to be the first point of contact when something breaks. Also, when you create a single device with an operating system that hosts other applications, then you become a single point of failure for those applications. If an application breaks your environment, or your environment breaks the application, the result for the customer is the same.
This creates a potential skill gap on both the SP and customer side that could lead to troublesome deployments and failures in support. With UC Today reporting that IT prosalready report that they don’t have the skills for the current digital transformation to SD WAN, uCPE may be a step too far for the industry until it upskills its employees. But also creating an opportunity for service providers who have the skills to diversify and provide more solutions to customers.
Also, I’ll be slightly self-serving here and remind readers that in addition to our research arm, Cavell does have a professional services division with many skilled network engineers that can help with tricky deployments and have been involved in multiple SD WAN rollouts.
The Challenge of Flexibility for Future Business Models
As a customer, the ability to easily pick from an app store of multiple formerly physical applications like CCTV management, Firewalls, or VoiP and switch on a virtual version of them on and off with zero network hardware change would be fantastic but we haven’t reached that stage yet. There are some hurdles like cost of hardware and application integration that must be surmounted first. There is also potential for interesting technologies to be developed such as machine learning based network mapping and configuration so installation of a new application can be truly automatic and zero touch. “Siri can you please install my new firewall and make it block Alexa across all company networks please.”
If we reach that point then customers will surely benefit, however, it may also require changes to business models such as an increased expectation of flexible switching – similar to what carriers are now facing where customers expect to be able to switch networks with a text. If a customer can download an alternative firewall product and online it with a few clicks, will they continue to accept long term contracts? Will your failure response have to be quicker, or risk losing a customer? What incentives can you offer in order to lock in a service beyond a month’s subscription found on most app stores?
It is also worth thinking about how this might change the sales process? Would increased flexibility and ease-of-install lead to more trial periods where different technologies are tested to find the best fit for the network. Or perhaps it will create more demand for options to break contracts if the service (as seen through the increased analytics of SD WAN) isn’t operating as expected. All of these questions are quite interesting but ultimately there is no solid answer until we get the technology to make it practical, but worth thinking about in terms of how the landscape may change in the future.
However, returning to the present, though these ideas are still quite far away, uCPE in some form is not. On the Colt webinar, they were happy to announce that this technology is moving away from being a concept with a few tests of the initial uCPE platform being trialled by some customers. Colt also mentioned that there may be a deployment announced shortly.
We look forward to hearing more from Colt about these deployments as they progress further. It will be especially interesting to hear what the main driver for the deployment was, whether it has enabled new business operations, or was it a legacy upgrade. My prediction is that it will be for a company that needed the flexibility of uCPE to do something special that legacy tech couldn’t manage well enough to make it viable, but let’s wait and see!
If you’re interested you can find the full webinar here – https://information.colt.net/ucpe-revolutionising-the-network-edge-with-colt