TL;DR — Key Takeaways

  • Microsoft and AWS are collaborating on a new networking service designed to simplify private connectivity between Azure and AWS environments.
  • The service combines Azure Multicloud Interconnect with AWS Interconnect – multicloud and will support dedicated links of up to 100 Gbps.
  • Open API specifications are intended to reduce the manual provisioning, coordination and lifecycle management typically required for multicloud networking.

Call it a milestone: Microsoft and Amazon Web Services are working together on a new networking service that will make it easier for enterprises to privately connect workloads running across their rival cloud platforms.

The agreement is a surprising shift in how these two dominant cloud vendors approach multicloud. The competition between the two rivals has been fierce, and they have not stressed interoperability between their respective platforms.

But times have changed. There has long been some customer demand for connection between multicloud environments. Enterprises typically place applications and data with different providers based on cost, technical requirements or access to specific services. And moving data and supporting applications across those environments can create another layer of network complexity.

Now, AI is adding pressure to solve this problem. Training and inference require large volumes of data stored in different environments, while enterprise AI deployments need predictable performance and security across platforms.

The new Microsoft-AWS collaboration directly addresses these difficulties.

The offering combines Azure Multicloud Interconnect with the AWS Interconnect multicloud service, creating dedicated links between Azure and AWS with bandwidth of up to 100 Gbps. The two companies are using open API specifications for network interoperability, which replaces much of the manual work needed to connect separate cloud environments.

For cloud customers, the key benefit is simpler multicloud networking. Connecting infrastructure across cloud providers can require IT teams to coordinate physical connections, synchronize provisioning and handle monitoring and lifecycle management across separate platforms. Microsoft says that process can require weeks or months.

Azure Multicloud Interconnect is designed to shift much of that work behind a cloud-based provisioning layer. Customers will be able to establish dedicated private connections through a simplified interface that hides much of the underlying network complexity.

Interoperability Beyond Azure-AWS

The project also extends beyond a bilateral Azure-AWS connection. The open specifications underlying the service were previously used for interoperability between AWS and Google Cloud. AWS says its Interconnect multicloud offering now covers connections with Azure, Google Cloud and Oracle Cloud Infrastructure.

The service is currently in preview, so its features and performance targets could change before general release. Microsoft said connections will support speeds of up to 100 Gbps at general availability, with capacity designed to expand as customers need more bandwidth.

AWS said the joint networking approach will provide MACsec security and target 99.99% availability.

The longer-term direction is likely more important than the initial 100 Gbps links. Cloud providers have traditionally competed by building largely self-contained ecosystems. Standardized networking APIs could make infrastructure placement less dependent on cloud boundaries, allowing customers more freedom to distribute applications and data while managing the environment as a connected network.

Frequently Asked Questions

What are Microsoft and AWS launching?
They are working together on a private multicloud networking service that connects workloads running in Microsoft Azure and Amazon Web Services.
How fast will the connections be?
Microsoft says the service will support bandwidth of up to 100 Gbps at general availability.
What problem does the service address?
It is designed to simplify the process of connecting infrastructure across multiple clouds, reducing the need for manual provisioning, physical coordination and separate network management.