Cisco’s got a stronghold in enterprise data center networking. Its presence in cloud providers’ data centers, however, isn’t as large as Cisco would like it to be.

During a tech talk today with Goldman Sachs, Cisco’s Ish Limkakeng, SVP, product management, data center networking group, discussed what the networking giant is doing to grow this segment of its business.

The company classifies cloud provider customers in two segments: major public cloud providers and software-as-a-service (SaaS) providers, Limkakeng said.

“With the public cloud space, we’ve been underexposed, and we view that as a big opportunity going forward,” he noted, adding that the SaaS provider space is a growing segment for Cisco as well.

Looking at the hyperscaler market, Cisco has the technology to “fully compete in every part of any given public cloud providers’ network, whether it’s in the core of the network or the data center spine and leaf tier, all the way out to the edge and compute,” he said, pointing to the Cisco-Google partnership that took center stage at last week’s Cisco Live event.

To boost its cloud-provider appeal, Cisco is changing its approach to how it works with these customers. “At the end of the day, they are doing things at scale that have never been done before,” Limkakeng said. “Ultimately it is a lot of hardcore engineering problem solving that has to happen. We’ve changed our engagement model to be more of a direct engineering conversation than a traditional sales and go-to-market and business-development exercise. That is how these cloud providers want to work with us.”

Cloud providers also want open networking and automation, he added. To that end, “we have opened up the model with which they can consume the technology, whether it is our hardware independent of our software, our software independent of our hardware, co-development, or partnering loosely with third-party providers.”

As an example of how the company is changing its approach, Limkakeng cited working with an unnamed cloud customer to improve core routing protocols.

For an enterprise customer, saving a few seconds in spinning up a rack wouldn’t be a big deal, he said: “It doesn’t make that much difference in a classic data center for the application.”

But for cloud providers, with tens of thousands of hundreds of thousands of severs in their data centers, shaving a few seconds off of this time is a big deal.

“In a mega-scale environment, that couple of seconds gets multiplied by tens of thousands of servers, and you have a business-impacting change,” he said. “Bringing up a rack in X amount of time faster — that’s not like saying we’re going to build a new switch or build a new product. It’s an engineering exercise around that problem, not a cleanly defined product."

Speed and agility are also paramount to cloud providers. But responding quickly and nimbly to customer isn’t always easy for a behemoth like Cisco. “With hyperscalers in particular, that change in engagement model allows us to experience those problem as our customers are experiencing them. It gives us a way of shortening the cycle to help with a solution.”