Why the Cloud Will Drive the Next Wave of Innovation in Networking

“Hi Network. Meet Cloud!” is our favorite tagline at Lavelle Networks, because it highlights the renaissance of innovation in data networking that is happening all around us, because the cloud demands it.

Let me explain. The definition of cloud is IT infrastructure (from simple virtual machines to AWS Lambda services) available on demand, but accessible only through a NETWORK.

While computing and storage technologies have undergone large scale rapid innovation in order to be able to create the huge data centers or cloud regions that we are now familiar with, we believe that the next wave of IT infrastructure innovation is going to be in networking.

The better, faster, safer our networks are, the more we will be able to grow the phenomenon of ‘Cloud’.

So what lies ahead? Where will we be 5–10 years from now in data networking?

1. TCP — For how long are we going to optimize it?

Anyone who has been in the data networking industry long enough, is probably disillusioned from the umpteen efforts to optimize the workhorse protocol of modern communication — TCP.

When it was invented 40+ years ago, networks were unreliable, frequently congested, and faulty. However, a vast majority of network software is built on TCP, and we have had waves of products optimizing this protocol using devices, software and patches.

Looking at the recent work Google is doing with QUIC or Apple with MP-TCP, it is evident that we are all looking for the next generation transport protocol, which comes out of the box “optimized” for modern, high speed, low loss, congestion-controlled networks.

It is time to create the next standard. Like ATM gave way to Ethernet, and how MPLS is now giving way to SD-WAN, it is important to innovate at the most fundamental layer of networking — TCP.

2. Wireless — Breaking the multi-Gigabit speed barrier

The latest 802.11 WiFi and 5G standards are here, and they are pushing wireless transmission to the Gigabit speed barrier. The real seed of this change began once we moved baseband processing to software, and we were able to use a high bandwidth medium like fiber between the radio and the baseband (CIPRI interface).

Wireless has never looked back since then. Wired local networks are almost going away, except within data centers where terabits of networking capacity are needed.

We will see transformative changes in metro wireless and long-distance wireless (like this high-speed satellite communication project — ISRO high speed satellite). New-age economies like India, China, Brazil have huge demands for bandwidth which only the multi-gigabit wireless technologies can meet.

3. Text — The next frontier for high speed packet processing

Networking packet processing traditionally has relied on packet formats which are fundamentally encoded in binary form, and not in text form. This helped make processing fast, less error prone, and more efficient on the wire.

But now this has changed completely. No network processing engine is complete without the ability to process text. As all application awareness is over HTTP or equivalent protocols, the need to do high speed text parsing is pervasive. API calls, application messages, media protocols are all encoded in different kinds of markup text languages, and we will see high speed text parsing engines being built in FPGA or Network Interface Cards.

The network can no longer be a transport, it has to be user aware, content aware and application aware. All of this just means loads of more high-speed text packet processing.

4. IPv6 — The silent transition

This is the most under-rated technology transition of all time. The first generation protocol for IP networking — IPv4, had a crippling limitation in that only 4 billion endpoints could communicate using it’s 32-bit addressing model.

Enter IPv6. This was a technology transition which was supposed to happen for years, and eventually when it happened, many didn’t notice because the network still worked the same way, but it could handle everyone and everyTHING (IoT) on the planet.

In the coming years, we will see this become the mainstream protocol in a visible manner, from it’s current invisible status.

5. Security — The perimeter is now in the cloud, not on your premise

Network security is still deployed from a physical premise mindset — IT teams deploy firewalls and security appliances literally at the edge of the network like the WAN, DC edge, DMZ.

But we now know that is not sufficient, because attacks originate in the cloud, permeate through the cloud, and are targeted either at end points (devices, hosts, phones) or at your virtual server and data assets in the cloud.

The future lies in securing your virtual perimeter in the cloud; you have to deploy a virtual WAN in the cloud, so that all of your users go through that virtual perimeter before touching your cloud assets, and vice versa, and nothing can get to your end points without going through that virtual perimeter.

Because the cloud is elastic by nature, a security perimeter in the cloud can expand and shrink based on the volume or density of attack traffic, unlike physical premise infrastructure which just crumbles under an attack today.

Technologies like SDP (Software Defined Perimeter — Cloud Security Alliance) and Google’s BeyondCorp are a glimpse of what will happen. With more than 100+ public cloud regions, any enterprise can create a 100-PoP security perimeter, with the right platform and tools.

6. SDN/NFV — Enterprises will lead the adoption

Did you really think this article would end without mentioning the buzzwords in modern networking literature — SDN/NFV?

While both of these technology platforms had their genesis in web scale data centers, and are currently the darling lexicon of Telcos and ISPs, the real future for this lies in enterprise.

The key reason being that the promise of SDN/NFV is actually operational efficiency, cost reduction, agility — all of those things which enterprises love, and now the platforms for enterprise adoption are ready. SD-WAN’s rising popularity is one such example, as it is the first application of SDN/NFV for enterprise.

Telcos and ISPs need technologies and tools to generate new revenue streams other than bit-pipe services, and the disillusionment regarding SDN/NFV is precisely due to that. Implementing SDN/NFV helps with reducing costs, but not necessarily adding new services.

Customers (enterprise or consumers) still look at the Telco/ISP for bandwidth — plentiful, low cost, high reliability, and want them to stay out of the way of their video content and application indulgences. The challenge with the content and application game, is that it is fundamentally OTT in nature.

Enterprises on the other hand, are racing into digital transformation journeys, cloud migration, omni-channel experiences, and desperately need their network infrastructure to be simpler, faster, safer, and will lead the charge to adopt SDN/NFV in their network.

That is what a peek into the future of Networking shows us. It might sound like a lot of change, and as an industry we have always been impeded by the need to keep older networks and protocols working, but we now live in an age where operating systems and applications are updated almost every week over the air, across the world.

For the first time, the drivers of communication are not client and server machines, but the billions of handheld devices communicating with hundreds of thousands of microservices-driven applications. And this world of mobile-first, does not fear change, because software updates are part of this new paradigm.

And therein lies the key to the future of networking. We will now be able to make rapid, tremendously innovative changes to make our networks what they need to be.

The Future of Networking is exciting, fast and it’s all software. So remember you heard it first, here — Hi Network. Meet Cloud!

This article was written by Shyamal Kumar, Founder and CEO, Lavelle Networks.