You also likely understand that our Glass Core™ is Fiber Mountain’s™ solution to the problem I mentioned above. Glass Core replaces many of the core and aggregation switches that are currently occupying space, drawing immense power and causing latency inside data center networks with hundreds of intelligent fiber cables that can connect any two devices on the network, creating what we call connectivity virtualization.
What you may not know, however, is that Fiber Mountain’s solutions can do more than just help you reduce cost and increase capacity right now; they can also prepare you for a future that we believe will look quite different from today. What does that mean?
Well, I look at it two ways:
1. Bandwidth Needs Are Speeding Up, Not Slowing Down: The need for more bandwidth seems nowhere near a plateau when you account for: the growing number of complex applications being used today; the increase in cloud deployments; adoption of real-time data analytics solutions for the Internet of Things; popularity of VOD services like Netflix and a host of other factors. As a result, investment in high-speed ports is skyrocketing, reaching $39 billion in 2014 according to a report from Infonetics. That same report notes that 40G is increasingly common in data centers, with 100G becoming the popular choice in the core. That means that data centers that are already struggling to support bandwidth demands, while also keeping costs down, are in for more difficulty moving forward if they do not make changes to their network architecture. These organizations will have no choice but to introduce even larger, more expensive power-hungry switches into their environments as they continue to scale.
2. The Advent of Silicon Photonics and Hardware Switching in Servers: Servers equipped with silicon photonics on board, along with hardware-based packet switching, will hit the market in 2015, and they will become more popular over the next several years as data center managers look to cut latency, reduce cost and simplify their networks. At Fiber Mountain, we believe the logical endpoint for this trend is that the server will consume a majority of the switching functionality at some point in the future. Our solutions allow data centers to prepare for that eventuality, as these servers with many cores and silicon photonics will be able to communicate directly with other servers, storage, and other destinations with multiple lanes in our Glass Core.
While I do not possess a crystal ball, I believe strongly that the aforementioned ongoing developments paint a clear picture of what the future of network architecture will have to look like. The old three-tier design in which we punt packets to larger and more expensive switches is simply not sustainable from a bandwidth, cost, performance and even environmental standpoint.
With that in mind, data centers are left with two choices:
A. Start preparing for the future now by adopting a simpler, less expensive and more powerful network architecture that allows them to scale easily
B. Wait until they are at critical mass to make a change
Which option will you choose?