Why Multivendor is the New Gold Standard for Data Centers


As the global economy continues to move towards what appears to be a cyclical downturn, IT teams everywhere will find familiar refrains of controlling costs resurfacing, with many turning to new technologies, including automation. But what if automation alone is not enough? Mike Bushong, VP of data center product management at Juniper Networks, explores what multivendor data centers have to offer in our volatile landscape.

As the global economy continues to move towards what appears to be a cyclical downturn, IT teams everywhere will find familiar refrains of controlling costs resurfacing. Of course, managing costs is nothing new, especially for IT teams whose primary function is to enable other parts of the business. In fact, containing expense has been a major driver of the industry’s automation ambitions for more than a decade. 

But what if automation alone is not enough? What if it isn’t even the leading lever that IT leaders and architects ought to be looking toward? 

Enterprise Management Associates (EMA) surveyedOpens a new window IT leaders across the United States and Europe and found that, while automation is helping to simplify processes, containing expenses is a different story. Nearly half of those surveyed said they use two different network automation tools, and more than a third use three or more. This overlap in solutions can increase operational complexity and result in budget bloat, even as the solutions themselves promise to reduce both factors. The push for efficiency has to focus on more than just automation and instead on choosing the right tool for the job.

Success Hinges on Having Multiple Stakeholders

For the cost-conscious, the primary economic lever—across all industries, by the way—has always been competition. In data center networking, the reason that merchant silicon is so important is not because it is somehow inherently less expensive than its proprietary alternatives. Indeed, having yet another layer in the supply chain means there are more companies that need to be commercially successful. 

The real value of merchant silicon is that it creates equivalency across a whole collection of product offerings. If the industry effectively standardizes device design around a set of common components, it not only allows but promotes more fierce competition. This has changed market dynamics completely. The beneficiaries? Anyone buying data center networking equipment.

But it turns out that leveraging common components is not enough to fully unlock the power of competition.

See More: Achieving Good Data and Database Modernization To Enable Successful Strategies

Data Center Operations Need to Embrace Choice

The key to competition is the notion of interchangeability. If parts can be swapped out with relatively little effort, then the competition is fair, and pricing dynamics follow as expected. When the barrier to change is high, the threat of replacement is lower, which naturally dampens the competitive environment. 

In data center networking, the barrier to change has historically been high. This is not because of the equipment but rather the operations that surround it. Change is difficult when tools and teams are oriented around a single supplier. While devices can easily be substituted for one another, retraining teams, reintegrating tools and updating processes is significantly more difficult. Effectively, a company’s operational stance ends up negating the economic benefits it might otherwise accrue.

We need to look no further than the current supply chain challenges. Companies everywhere are grappling with long lead times on hardware. In the consumer space, customers would merely shop elsewhere if a store did not have something in stock. But if data center operations restrict choice, there is no elsewhere. The result? Companies are in a tenuous position as they wait for equipment. Even worse, they have to pay more for that equipment because they are competing with so much pent-up demand. It’s literally the opposite of the market dynamics most expect.

Multivendor Designs Offer Critical Flexibility

The situation might appear bleak, but emerging architectural best practices will provide long-term relief.

The hyperscalers have been multivendor out of necessity. They simply cannot rely entirely on a single supplier. The purchase volumes are too great, and the need for capacity is too acute to create vendor dependency. As a result, they built from the ground up with multivendor as a requirement. Of course, they have legions of software developers to solve their operational lock-in problems, but the point remains: multivendor designs provide flexibility, which unlocks economic leverage. 

But how do enterprises unlock similar value? 

There are tools on the market today built expressly for multivendor management. By abstracting control, these tools act as translators of network intent into the on-device behavior that makes it possible. This is a natural extension of the software-defined networking movement that started in the late 2000s. The notion of over-the-top management created an architectural separation between the point of control and the underlying devices. While the first generation of vendor-provided tools merely moved management to some centralized software, more modern tools have arrived that provide a shim layer between the operations and the underlying equipment. 

While today’s supply chain woes did not drive the initial development of this class of tooling, they do underscore its importance. Imagine being able to put out an RFP based solely on the switching requirements and allowing vendors to compete for the business. It would shift control from the equipment manufacturers back to the users. IT leaders could choose based on price, performance or even lead times. And if the choice was transparent to the team required to deploy and run the network, whatever benefits emerged would not be offset by some change overhead.

Intentionally creating more competitive bidding environments eventually leads to vendor specialization. That is to say that a vendor that cannot win purely on price might choose to lead with performance or ease of use, or supply. When this happens, it sets a new bar for a particular axis, which will naturally cause others to change their view of what “good enough” for a particular characteristic means. Essentially, this reinvigorates innovative drive, which should bring broad benefits to customers regardless of what they are optimizing for.

It’s not just about the supply chain or economic downturn, though. The current work-from-home movement has expanded the hiring boundaries of the largest cloud providers and technology companies. Where their recruiting used to be confined to areas like Silicon Valley, they can now hire in Des Moines. This is going to create recruitment and retention issues for companies everywhere. Combined with a dwindling supply of practiced networking professionals, it means that experience will become more scarce and more expensive. 

If multivendor architectures allow operations to be separate from the underlying device-specific command line syntax, it opens up the hiring pool immensely. And if those solutions also succeed at simplifying the Day 0, Day 1 and Day 2 operational tasks, it means that companies can effectively run expert data centers without necessarily needing a full complement of data center experts. 

How are you employing multivendor architectures for your data centers? Share with us on FacebookOpens a new window , TwitterOpens a new window , and LinkedInOpens a new window .

Image Source: Shutterstock