Cloud technologies have revolutionized the IT industry, but the process for introducing a new IT application hasn’t always been as easy as it is today. In the past, every project, no matter how large, demanded the involvement of multiple teams and a project plan with a big Gantt chart, and IT teams not only had to deploy the application for use within the enterprise but also had to build and configure the servers that the application was to run on. Now a single person can implement a new application in a couple of hours with some code and a credit card.
So, what changed?
Separating the applications software from the hardware has enabled these two elements to innovate and scale on different paths. Hardware has moved to volume-based, low cost equipment, inside shared, very high scale, automated and people-free data centers, which enables software to iterate with its one velocity separate from any infrastructure constraints, which can be deployed on demand.
The telecom network world is still mostly stuck in the old IT world. The complex processes of designing, planning, building and configuring networks are undertaken in small, incremental steps because the cost to do so otherwise, including equipment and people, is too prohibitive. Moreover, the process of building telecom networks is divorced from customer demand and instead, is done based on forecast demand. Even worse, when enterprise customers request telecom connectivity the process of ordering, designing, building and connecting the enterprise can sometimes take months.
Being a network guy myself, I can say it’s true that networks are more complicated than IT, but fundamentally, both require the same solution: the separation of software from hardware, and a shift towards volume-based, high capacity hardware. The IT industry has created cloud-based capabilities with most of the pieces needed in the WAN (Wide Area Network) with one big exception: the ability to scale out.
The ability to scale out remains a challenge because in addition to adopting cloud technologies and principles, the concept must also be applied to location awareness. This adoption has been the focus of leading telecom operators and recently received a large boost with the announcement that OpenStack’s 16th (and latest) release, codenamed Pike. Pike puts an emphasis on composable infrastructure, which is stated to “make possible use cases like edge computing and NFV.” But this effort is wasted unless the WAN connectivity also moves away from the complex incremental world to the same cloud principles of separate software configuration and control, and high volume, high scale, low cost physical paths.
To achieve cloud networks, we need to:
- Separate software from hardware
- Move to high volume, low cost hardware
- Execute on the above in thousands of mini cloud locations
- Enable these mini clouds with programmable and flexible high capacity connectivity
There is much essential work underway in open communities such as Facebook-sponsored Telcom Infra Project, and the ETSI MEC ISG. Connecting thousands to mini cloud locations with high capacity and highly reliable connectivity is also essential. Optic fiber offers an opportunity; however, it is expensive to deploy and isn’t suitable everywhere. A viable wireless solution that solves the scale-out issue in the WAN is needed.
The Collinear Hybrid Wireless solution offers a rapid path to the distributed cloud network. It is a high availability, very large-scale solution that is externally controllable, with open standard interfaces to solve the distributed cloud connection bottleneck.