Inside Advanced Scale Challenges|Monday, November 20, 2017
  • Subscribe to EnterpriseTech Weekly Updates: Subscribe by email

Why Cloud Wireless Is the Fastest Growing Network IT Segment 

(Supphachai Salaeman/Shutterstock)

Many companies are moving key applications (CRM, HR, finance) to the cloud to maximize IT efficiency, minimize IT costs and improve business agility. For the same reasons, key infrastructure elements (security, storage) are also going to the cloud. However, wireless networks have been slower to adopt this transition, with the vast majority of the wireless LAN (WLAN) market still delivered via on-premises controllers.

Moving wireless to the cloud gives CIOs a more scalable and resilient infrastructure with better operational simplicity. In addition, it gives CMOs and business owners actionable insight from the petabytes of data flowing through wireless networks today. This is why cloud wireless is the fastest growing segment of network IT, with one-third of the total market expected to transition to the cloud by 2020, according to IDC.

The first generation of cloud Wi-Fi products, introduced in 2007, lacked the agility and scale to satisfy today’s requirements, and they can’t address key WLAN requirements, such as automation using machine learning and visibility into the mobile user experience.

However, cloud technologies have evolved a lot in the last 10 years, bringing more resiliency and performance.  In addition, AI and big data analytics are changing the picture, bringing true web scale and agility to the burgeoning cloud WLAN market. When all these advances are combined together, the new generation of cloud WLANs not only simplify WLAN deployment, they also automate ongoing operations for substantial cost savings. This has allowed the modern WLAN to shift its focus from the infrastructure to the user, enabling Wi-Fi to be delivered as a predictable, reliable and measurable service.

Amazon, Google, Facebook, LinkedIn and other digital leaders, which must correlate massive amounts of information using machine learning in the cloud, provide a great model for harnessing those same principles to build user-centric wireless networks.

These emerging WLAN platforms are built on a microservices architecture using the latest cloud, AI and wireless technologies and can provide capabilities not available on first generation cloud WLANs, such as: rapid deployment of new services without impacting existing services; reliability due to service containerization; elastic scale; and actionable insight using global datasets and machine learning.

Traditional wireless architectures fail to deliver at scale due to monolithic designs that use vertically integrated systems. Take, for example, a distributed enterprise spread globally across 3,000 locations with 100,000 access points. In the old world, on-premises or cloud-hosted controller pairs would have to be replicated for resiliency. In certain scenarios, additional controllers would be required for providing management and troubleshooting, as packet captures, debug logs and event logs eat up substantial processing power.  On top of this, some industries (e.g. retail and hospitality) require even more controllers to be pre-instantiated to handle temporal surges in demand.

Here is where a microservices architecture has an advantage. It can enable administrators to monitor utilization of different services, and scale each module up or down dynamically without requiring end-user intervention. That’s elastic scale, without a physical cap on the number of access points, client devices or sites (per customer or globally.)

Shrinking and expanding resources with minimal lead time is a fundamental advantage of being built on the modern cloud, as the network IT team no longer needs to worry before a critical event (such as Black Friday for retail stores) as to whether pre-instantiated resources will scale and perform to the demands on the network.

It is worth noting that elastic scale is highly dependent on load prediction. While the load coming from each individual customer can significantly change in a short period of time, the overall aggregated load across many customers changes much more slowly.  Hence, having a unified production environment for all customers is much more desirable than first-generation cloud solutions where separate “shards” are set up in the cloud for each individual customer that is less prone to scale dynamically with system changes.

Neural networks and machine learning concepts have been around since the 1950s.  However, the compute power needed to solve large computational problems did not exist until the modern cloud emerged in the mid-2000’s. Cloud infrastructures like AWS, Google Cloud and Azure have turned AI into a mass market technology by pricing compute cost-effectively (with elastic growth), enabling companies of all sizes to quickly build AI platforms on massively scalable and secure global cloud infrastructures. WLAN platforms built using AI can automate Wi-Fi operations, simplify troubleshooting, detect anomalies, analyze trends and provide predictive recommendations.

Security is another important consideration. We live in a world where data center breaches are in the headlines almost monthly, and there is a fear of security in public cloud due to fears around data security. Clearly, there are myths that cloud computing is inherently less secure than traditional approaches. The paranoia is due largely to the fact that the approach itself feels insecure, with your data stored on servers and systems you don't own or control.

The truth, however, is that public cloud is often more secure than most traditional data centers.  That is because cloud providers have better security mechanisms in place and are more paranoid and attentive to security risks throughout their entire stack. The cloud providers are much better at systemic security services, such as looking out for attacks using pattern matching and AI technologies.  In addition, they are always leveraging the most up-to-date security technologies and solutions for minimizing exploits. The next-generation cloud WLAN takes security even further with hardened servers, highly restricted user access, industry-standard encryption at various levels, obfuscation of user information stored in the cloud, and more.

Outdated WLAN infrastructures cannot meet the needs of the modern enterprise. The move to the cloud was a great first step, but first generation cloud architectures lack the scale, resiliency, agility and elasticity needed for today’s business requirements. The next generation will bring Wi-Fi that is reliable, predictable and measurable, as well as easy to deploy and cost-effective to operate.

Bob Friday is CTO at Mist Systems, a provider of self-learning wireless networks using artificial intelligence.

 

Add a Comment

Share This