PacketFabric’s Anna Claiborne: Automation and Programmability are the Future of Cloud Connections
by Josh Anderson
SAN FRANCISCO, CA — CapRE Data Center Summits expend a lot of bandwidth on the challenges that firms face when migrating data to the Cloud. However, that are many layers to that process, since there are different kinds of cloud. In fact, even cloud-native companies face struggles when deploying hybrid cloud architecture. So at CapRE’s Northern California Data Center Summit, we gathered a panel of regional data center insiders to share lessons learned and perspective gleaned from this process. Moderator Mike Nguyen, Founder and CEO of Inflect put it well when he asked the panel — how can you convince cloud native companies that they need to do something more than just buy a MacBook Pro? Anna Claiborne, Senior Vice-President for Software Engineering at PacketFabric shared some interesting analysis.
“Connecting to the Cloud can cause quite a few issues,” she replied. “First you have your security and performance issues, which, just moving from a VPN for a direct connection can solve. So that’s easy to say – let’s just move over to a direct connection. But there are a lot of different layers just in that direct connection.”
Claiborne then stressed that cloud service providers are not the same thing as network service providers – far from it, in fact. “They know very well how to provide network services to themselves in their own internal environment, but they do not know how to provide services externally,” she explained. “The limitations of most direct connect products are that we’re going to give you a port or we’re going to give you a common internet pairing. That’s the extent of what they give you. There’s not a lot of help to set that up. You have to have that expertise in house, which we’ve established doesn’t always necessarily exist.”
So, according to Claiborne, the crux of the question lies a bit deeper. As an enterprise, or as an external cloud user, how do you make that network connectivity work, in order to move a workload from point A to point B? “That is where, from my perspective, the biggest challenge is,” Claiborne confirmed. “And especially when doing that, even doing that today — a lot of service providers are doing that through Amazon – it is still a very manual process,” she continued.
“There is nothing around that which exists to automate it, except for the newer platforms like ours. You can’t simply turn on a connection at will,” said Claiborne. “If you want to, it usually takes 30-60 days and after you make that kind of investment, you ‘re not just going to turn it off. So the only way to make these things truly dynamic is through automation and programmability. And to have APIs to do this sort of stuff.”
Claiborne then pointed out that this issue is going to get more critical as the data center industry heads into the future. “Because this hybrid environment that we talk about right now, it’s in-house data center vs public cloud,” she explained. “That’s going to become the edge very quickly. So you’re adding in another layer of complexity in terms of where you’re going to need to move content, where you need to move data, and where you need to move workloads. And unless you can do that quickly in an automated fashion, it’s going to become more complicated than it already is.”
Check out previous CapRE Insider Reports covering this panel: