For many organizations, multicloud simply means running different workloads on different clouds. It's a situation that can lead to data silos and operational inefficiency as the different clouds aren't often actually connected to each other. To help address that problem,  Google today is announcing a series of updates to improve multicloud networking with new Cross-Cloud Interconnect and Private Service Connect enhancements.

Cross-Cloud Interconnect is a new service that is generally available as of today that directly connects Google Cloud with Amazon Web Services (AWS), Microsoft Azure, and Oracle Cloud Infrastructure. With Cross-Cloud Interconnect organizations can establish secure high-bandwidth connections between other public cloud platforms and Google Cloud, enabling them to run applications on multiple clouds, host SaaS in a multicloud environment and migrate workloads between different cloud platforms. With the Private Secure Connect service, organizations can have a private, isolated connection into Google cloud that can now also extend into third-party services that Google is directly connected with.

“Networking is the foundational fabric that allows customers and organizations to move workloads into the cloud as well as modernize their applications,” Muninder Sambi, VP and GM, networking and security at Google Cloud, told SDxCentral.

Multicloud networking beyond Google Cloud

Google Cloud Platform has a large footprint spread across 35 regions around the world. Even with that large deployment, Sambi noted that Google is well aware that organizations are increasingly adopting multicloud strategies.

[caption id="attachment_130736" align="alignnone" width="710"] An example of Cross-Cloud Interconnect between Google Cloud and a cloud service provider. (Source: Google)[/caption]

Cross-Cloud Interconnect is an offering designed to enable multicloud, such that organizations can manage and scale workloads running across different cloud providers.

“You can also host SaaS applications with multiple clouds or even migrate workloads from one cloud to another,” Sambi said.

Sambi explained that the Cross-Cloud Interconnect does not require any new customer hardware and can be provisioned by Google with bandwidth of 10 gigabits per second or 100 gigabits per second (Gbps). The interconnect with different providers is a tier one provider connection with the capability to provide logical networking overlays for workload management. Sambi noted that Cross-Cloud Interconnect is a managed service that will provide organizations visibility into overall cloud networking performance, including round-trip across cloud providers, latency and packet loss metrics.

The Cross-Cloud Interconnect has been in early access for several months, with some big name organizations, including Walmart.

“What Walmart was able to do is simplify connectivity between cloud providers, shorten the time to production and overall reduce costs,” Sambi said.

The Intersection of SD-WAN and Cross-Cloud Interconnect

As organizations move to multicloud deployments, they are also often moving to SD-WAN deployments as well.

Sambi noted that SD-WAN is all about branch aggregation to help improve application quality and experience as organizations connect users to the cloud. Once the users are connected to the cloud, the workload in the cloud could be distributed. Google can connect with SD-WAN providers via its Network Connectivity Center (NCC).

With Google's new multicloud networking updates, Sambi said that organizations can now connect via SD-WAN into Google Cloud, and then connect to multicloud applications in AWS, Azure or even OCI with low latency and high reliability.

BGP improvement land in Google for multicloud networking

As part of the networking update, Google has enhanced its Border Gateway Protocol (BGP) protocol support. BGP is widely used across the internet to help route traffic across different providers.

Among the updates to BGP, Google now supports custom learned routes, which is important in a multicloud environment. Sambi said that without the custom learned routes, when an organization is traversing the internet, it needs to have the public IP address and route to get to the right location.

Sambi added that when an organization’s data traffic is traversing the internet, there is a need to have a public IP address that provides a published route on the internet. With custom learned routes, an organization can now use private peering, so they don't need to advertise private addresses outside into the internet. As such a virtual private cloud (VPC) on one cloud provider platform can be directly connected to a VPC on another provider in a more seamless approach with lower latency.

“What we're trying to solve is making it really simple for customers to adopt multicloud and be able to integrate their workloads seamlessly between any of the cloud providers,” Sambi said.