Cloud Native ComputingDevelopersDevOpsEdge/IoTOpen SourceTo The Point

How Different Is Networking In Cloud vs Traditional Data Centers


Dan Spataro, Vice President Of Infrastructure Operations at Linode, joins me to talk about the difference between networking challenges between cloud-native computing and traditional data centers.

Guest:  Dan Spataro
Company: Linode
Show: Let’s Talk


So I think that’s the point of being a cloud infrastructure provider, right? We’re trying to abstract away as much complexity for the customer as possible. They just want to run their workloads run their workloads reliably, and not have to worry about getting into all the details behind how the cloud works, you know, that’s our job. And there’s tons of details that go into all that. So building a cloud network, obviously, it’s pretty complex. And there’s, there’s a bunch of different technologies that go into it. You know, I think the main technologies that I’d like to talk about are interconnection, you know, backbone in the data center network. As far as interconnection, you know, we put a ton of effort into understanding, you know, what external networks around the globe are important to our customers. Like, I personally fly around the world and go to all these peering conferences, meet with, you know, other networks to develop these peering relationships, and give our customers the best and most direct connectivity possible, which allows for the lowest latency, the least chance of variability due to the internet. So if we hand off a connection, you know, to these networks, or customers are getting direct connectivity, which is, you know, obviously far superior to having to ride, you know, unknown networks, unknown hops on routers. And then we also tile or our pops in our regions with a backbone. So we have a unified global ASN, customers can reliably transfer data between our regions, it stays on our backbone, which eliminates the variability of the internet also, which is awesome. And then, as far as the data center network, you know, it’s we have to be able to scale you know, hundreds and hundreds of 1000s of instances, provide features and all that is done, you know, using various overlays and an Open Source technologies.