Have an account?

  •   Personalized content
  •   Your products and support

Need an account?

Create an account
AI-enabled hyperconverged infrastructure bolsters enterprise goals

AI-enabled hyperconverged infrastructure bolsters enterprise goals

Scott Robinson

by Scott Robinson

Hyperconverged Infrastructure – fully virtualized servers, storage and networks – boosts enterprise IT efficiency and performance. AI-enabled hyperconverged infrastructure will take these benefits to a new level.

Hyperconverged infrastructure promised to bring customers simplicity and one throat to choke. But has this architecture delivered?

Hyperconverged infrastructure packs a punch with its technology. It brings the virtualization of servers, data storage and networking together in one box. It is designed to enable companies to manage their IT infrastructure in one place and from one vendor. That promises ease of management and simplicity, but it also poses challenges. Let’s take a look at how HCI has evolved and whether it’s better positioned today to serve customer needs.

Putting entire enterprise systems under the virtualization roof opens possibilities. When data storage is dynamically reconfigured, the network serving it can be simultaneously optimized for that reconfiguration; when inbound data traffic greatly increases, even an enormously complex system can self-throttle with dynamic, just-in-time buffering. New technologies such as edge computing—which IDC predicts will be an architecture for 20% of enterprises by 2021—become easier to implement; and the implementation of inter-system messaging can finally become a no-touch endeavor.

Put another way, the cost and maintenance of all the physical considerations we put into network and data storage planning for efficiency and scalability (to say nothing of the time we spend implementing them) can be dramatically reduced with HCI.

Evolving at the speed of business

For some time, we’ve had self-optimizing data storage and dynamic networking. At the same time, virtual machine and application programming interface (API) technologies have evolved, and artificial intelligence (AI) has matured as well.

We’re now on the precipice of an AI-enabled business paradigm shift, as more and more decision-making processes and workflows are handed over to data-driven automated services.

Over the past decade, the phenomenon we’ve noticed, as cloud technology has become the centerpiece of enterprise IT, is that these technologies are increasingly interdependent in their evolution. The more they intertwine, the more each becomes a driver in the development of the others. Cloud computing enables the Internet of Things (IoT); the explosion of IoT necessitates edge computing; and now new network infrastructure pushes HCI, which in turn gives a push forward to them all.

What will this look like?

AI-enabled hyperconverged infrastructure, by industry

An AI-enabled HCI environment in the enterprise, under AI control, could be powerful.

AI is primarily about finding and responding to patterns, and patterns are present in so many aspects of IT infrastructure. An AI-empowered HCI environment can accomplish several goals:

  • Automated archiving of data. AI plus HCI will provide spontaneous scaling of archive storage and corresponding downsizing of transactional resources, resulting in zero-sum resource consumption.
  • Self-reconfiguring of networks is commonplace. AI-enabled hyperconverged infrastructure will enable no-touch reconfiguration of independent networks into a single optimized network, for example.
  • Blockchain. AI-driven hyperconverged infrastructure can dynamically reconcile differing paradigms when two or more blockchains merge.

What will all of this mean for business? Here are some possibilities, in hypothetical examples:

Supply chain/logistics. With AI-enabled HCI, companies can spontaneously deploy trucks, drones or other transportation mechanisms as needed. If a cataclysmic disaster hits a coastal city, the local metropolitan network automatically analyzes physical disruptions and shortages in transportation and utilities, access to resources and supplies and emergency services. It then spins up edge computing resources to compensate for increased IoT traffic, publishing critical schemas and APIs for secure external access as it goes. IoT data is dynamically rechanneled to disaster response systems.

Put another way, an emergency response network springs to life automatically, self-optimizes, and is accessible to emergency services and commercial traffic within minutes. IoT traffic is rerouted, and relevant data sources preserved. This provides immediate disaster analyses, and decision support analytics become available virtually in real time.

Healthcare. The federal government passes sweeping healthcare legislation that requires an industry-wide retooling of enterprise systems, regardless of role, to accommodate new data storage, security and process standards. Plus, new data-formatting requirements and IoT accommodations come into play. (And these new standards must be implemented within months, not years.)

All healthcare providers, payers and services that are based in HCI can immediately reconfigure their internal data storage resources. Networks are passively reconfigured to securely merge with partner networks. These would be optimized according to anticipated data traffic patterns already analyzed in AI simulation, based on extrapolation from previous patterns. Blockchain security for separate networks (transporting patient records) is reconciled according to shared requirements. And partner organizations can spin edge resources for increased IoT support on-the-fly. The healthcare enterprise has implemented, tested, and optimized for the new federal standards within days, not months.

Retail. Two national retail chains merge. Both have enterprise HCI environments, which makes their merger all the more practical; new data storage endpoints are generated to consolidate point-of-sale data into a single virtual storage array, with reconciled schemas so that both transactional processing and analytics can proceed uninterrupted, without elaborate extraction and transformation. Databricks spin up automatically to integrate the marketing analytics of the two chains, so they can continue to function until more permanent processes are decided upon. Also, the two point-of-sale networks, as well as the analytics-supporting IoT/edge networks, are merged, reconfigured and optimized by AI. The two retail chains are unified within a couple of weeks.

Better than the sum of both

None of this will happen tomorrow. It may take months to play out. But all of the technology outlined above is already on the shelf; it simply hasn’t yet been integrated and tested for this dynamic.

What stands in the way? There are a few obstacles .

  • AI-enabled hyperconverged infrastructure may compound the resource problems that have been associated with traditional HCI systems. While an HCI system is optimized for a balance of resources, it may need more of a single resource for AI-driven HCI systems to work. Adding more CPU, storage or bandwidth adds costs and wastes resources.
  • Further HCI systems need power, often more than your traditional server setup might require. AI-enabled hyperconverged infrastructure is likely to exacerbate that problem.
  • An abstracted AI standard capable of cleanly performing pattern analysis and indicating the right action to take does not yet exist, and there isn’t a clear path to defining it, let alone achieving it.
  • A mix of HCI vendor platforms may create issues for AI-enabled hyperconverged infrastructure in terms of clean integration and the systems being able to talk to one another. 
  • Blockchain will have to co-evolve in harmony with HCI and AI technologies, in order for all three to play well together.

In short, there’s a lot of work to be done to make AI-enabled hyperconverged infrastructure a reality.

For more Cisco news:

For more Cisco products:

Scott Robinson

Scott Robinson is director of business intelligence at Lucina Health in Louisville, Ky.