After the SD WAN trial discussed in my previous article (you can find this here), Colt began a three-phase deployment which is expanded on below.
Phase 1 was to deploy the physical hardware to manage the uCPE (provided by ADVA). The piece of technology was designed to handle not only the current functions that needed to be virtualised but also with capacity to handle any other technology that might need to be virtualised in the lifecycle of the product.
This is an interesting choice because you must sell a customer on a piece of technology that they will not use the full functionality of for some time. This is nothing new in technology as ‘future-proofed’ capacity is both a good selling tactic and also a wise decision to prevent having to repeat expensive purchases too soon. However, due to the shifting nature of cloud services and applications, it can be hard to predict exactly what future capabilities may be needed.
This is the real stumbling block for the development of uCPE. The technology hasn’t reached a point where it is cheap and generic yet which means that you must pay for an expensive deployment. Part of this cost is that you must sell increased capacity, but no one can really predict what the future of enterprise virtualisation will look like yet. Determining how much capacity to buy/sell is a struggle with large consequences both for the provider and the customer if calculated wrong.
This makes it a very difficult selling prospect to enterprises who want solid predictions, coupled with lifecycle estimates and cost-benefit analysis. This isn’t to say that uCPE won’t suit some agile businesses who have the money and need the flexibility provided by the technology. These companies could use uCPE to enable new business models and ways of working which will justify the expense as a strategic purchase rather than just network replacement.
One interesting area that this technology could work, is selling to SME/SMB who have much more predictable growth and needs. These companies would be much more ready to accept a specific offering on uCPE without the increased capacity or the control to change and tinker with it. Service providers can sell these boxes in as a direct transformation upgrade, replacing three boxes with one that provides the same service and reduces costs.
Also, you have the challenge that the hardware a uCPE is replacing is tried and tested, and already installed. So, you are not only selling a replacement, you are trying to justify additional costs on top of replacing something that in most cases does the job adequately. What this technology needs is some good use cases for things that a legacy technology cannot do to justify the spend. If these develop, coupled with the inevitable lowering of the price and a bit more standardisation may see this technology become more widespread in a few years.
Phase 1 also included an application layer SD WAN treatment including tracking, analytics and optimisations, tracking what applications were being used and what worked best. This is likely to become a standard part of network transformation as SD WAN technology becomes more prevalent as optimisation can smooth over a lot of network issues across companies of any size.
For phase two the functions of the network were changed to be virtual, but all the old hardware was left running in paralell until it was guaranteed that the new system was working. The hardware was powered down one at a time until the new system was up and running.
This seems like a practical approach to SD WAN/uCPE deployments, especially when you are working with a new technology, or at a company that cannot afford to spend days with teething problems on a new network. However, it is also a sign of the lack of maturity and reliability of the technology because it isn’t trusted to make a direct switch without a backup/transition phase.
This was the removal of unneeded hardware and the determination if it could be reused for any future functions at the site or at an alternative location. This is a smart addition to any transformation process as it reduces cost and should be factored into the sales plan as it can help customers feel better about deploying new technologies.
– Simplified estate and deployment
– Reduced hardware footprint, comms room footprint and also utilities costs
– Changes in deployment and to the platform simplified drastically
– Management processes mirrored from old estate to new, meaning procedures are maintained or improved on
– Simplified contract management on the hardware, 1 device = easier to handle
In the next article in this series I’ll share some thoughts on the final part of the webinar where they discussed the future of uCPE, so look for that in a few days
You can find the webinar itself here – https://information.colt.net/ucpe-revolutionising-the-network-edge-with-colt