I had a chance to sit down with Cisco CIO and Senior Vice President Rebecca Jacoby and several other editors at a roundtable talk after her Interop keynote last week. For an executive at a multibillion-dollar, publicly-traded company (with all of the media reticence such a position might entail), Jacoby came across as open, genuine, and direct.
She spoke at length about Cisco’s internal IT operations, including the impact of SDN and why Cisco uses OpenStack.
I asked her about the kinds of complaints she gets about Cisco IT from internal customers. Her answer was immediate: “Nothing’s ever fast enough. When you work for a tech company, IT is a specific challenge.”
She noted that even though the internal IT team believes it’s at the cutting edge, “...the expectations of the business are far ahead of what we can deliver.”
“People will complain all day every day about their specific challenge,” said Jacoby. “The more personal it is, the more likely you get concerns.”
She has to balance user concerns with other IT responsibilities. “What the business doesn’t complain about, but would get us fired, is security or outages. The business never says to you ‘Did your resiliency get better?’ But that’s still our job.”
“You can never make everyone happy, so you have to decide where in your architecture you give people choices,” said Jacoby. To that end, IT keeps an eye on what kinds of tools employees use to see if it can make a recommendation to the organization as a whole.
“We just signed a deal with Box because that type of tech is used extensively and we want to lean people toward the technology that’s best from an enterprise risk perspective,” she said.
She also wants to provide more standardized resources in Cisco’s internal private cloud to make it easier for users and business units to get services such as compute and storage. “If I can automate it, it’s standard.”
Given all the industry talk around SDN and network programmability, another editor wanted to know about Cisco IT’s plans around ACI and APIC. Does the company eat its own dog food?
“We’re an alpha customer,” said Jacoby. “There will be a migration associated with it. I will have some things deployed within 90 days. In two years I’ll be running all our workloads on APIC.”
Jacoby said her office had interactions with ACI at the development phase. “I’ve had three or four guys working with it from the beginning. We in the CIO office know what’s coming and can say ‘This makes sense, this doesn’t.’”
While Jacoby will use ACI and APIC internally, adoption of Cisco products isn’t always a given. “We aren’t actually required to use the products, but we have a propensity to use the products,” she said. “We’ve had situations in the past where we didn’t want to use the products.”
In addition to Cisco gear, the IT organization uses a variety of third-party products and software to run its private cloud, including open source software. In particular Jacoby called out OpenStack, which Cisco uses as part of its orchestration layer.
“OpenStack appeals because it’s efficient and we think it’s going to scale,” said Jacoby. “It’s not totally about saving money. It’s about standards and flexibility. Orchestration management is a very challenging space. IT organizations historically have done a lot of customization because a vendor can’t identify all your specific needs. OpenStack uses the community to evolve the space and you can choose pieces to plug and play.”
Another editor asked whether network automation would enable Cisco to cut costs because it would require fewer people to run the network.
“We never have enough people,” said Jacoby. “The reality is, there’s not enough skill sets to cover this kind of work in the industry.” Instead, more automation would allow existing staff to take on new challenges.
“You have to help your workforce move forward,” she said. “You raise the level of what those people are doing where you still engage and use the great minds of your network and security guys.”
She did note that she expects operational costs to decrease because automation would cut down on mistakes. “There’s a percentage of operational outage incidents are related to human error, so we’ll see a reduction there,” she said.