AWS wants Microsoft Azure customers to use its migration service to shift virtual machine workloads to its cloud, amid a competitive landscape that's increasingly focused on multi- and hybrid cloud deployments.
The move extends AWS' existing Server Migration Service (SMS), which previously catered to on-premises VM workloads with support for VMware vSphere and Microsoft Hyper-V migrations.
Organizations can now use the agentless service to group together their Azure VMs and move applications over as a unit, rather than having to replicate individual servers or decompose application dependencies, AWS said.
It's also possible to automate replications of live server volumes, which is critical for larger-scale migration efforts, according to AWS. There is no charge for the Azure-to-AWS migration tool, but customers pay for AWS resources such as Elastic Block Store and EC2 instances once their VMs are live on AWS.
Cloud vendors race to expand migration capabilities
With the new Azure-to-AWS migration service, AWS is playing catch-up, of sorts, to Microsoft, given that the latter has positioned Azure Site Recovery as a tool for AWS-hosted VM migration projects since 2015.
But AWS wants to maintain and grow its market share lead over No. 2 Azure, and has made workload migration a priority in other ways, particularly with its January acquisition of CloudEndure.
AWS, well-known for its torrid pace of internal development, rarely purchases companies. But CloudEndure's disaster recovery, backup and migration tools apparently proved too tempting a prize. The CloudEndure software is sold by SaaS subscription and addresses more complex scenarios than AWS SMS.
Google also invested in migration tools with its May 2018 purchase of Velostrata. The startup had previously offered support for AWS, Azure and Google, but now its webpage simply states that it provides "fast, flexible and safe migration to Google Cloud."
Azure-to-AWS migration may tempt, but cost considerations loom
Most enterprises plan to deploy a multi-cloud strategy. As a result, the market has responded to this trend with a thicket of service, but it still hasn't made the process any easier. The cloud vendor price wars of several years ago have ebbed of late, and combined with the ever-larger range of SKUs available from each provider, cost calculations around migrations get more complicated.
An enterprise might want to migrate workloads between clouds for myriad reasons, including better contract terms, a deteriorating relationship with an existing provider or a corporate merger that brings with it a large IT footprint hosted on other platforms.
Still, just because the options are plentiful, it doesn't make doing so a snap decision. IT executives need to use caution when there's no commercial long-term guarantees, said Holger Mueller, an analyst at Constellation Research, in Cupertino, Calif.
"All cloud vendors make it easy to move in, but very hard or even impossible -- less so technically than commercially -- to move out," he said.
For example, cloud egress fees -- money charged by a provider when a customer moves data out of its systems -- impact both initial migrations and decisions on whether to reverse a migration down the line.
And egress fees are just one factor associated with cloud lock-in, which more companies will try to avoid as they deliberately take a multi-cloud approach to their IT investments.
Ryan MarshDevOps and serverless coach, TheStack.io
At some point, growth in the cloud market will stall as the major cloud providers will reach the total addressable market, said Ryan Marsh, a DevOps and serverless coach at consultancy TheStack.io, in Houston. Creative ways to ease onboarding, which subsequently increase churn for competitors, will become essential for cloud providers.
"I would hope to see Azure continue to offer more onramps for AWS customers and vice versa," he said.
But even if migration between clouds becomes easier, such moves hardly guarantee parity between cloud environments, Marsh said. "What many people don't realize is that the cloud vendors are already more different than they are alike with regard to APIs."
User sign-on and identity management is one good example of where difficulties can crop up, according to Marsh. "Try transitioning a workload from AWS Cognito to Azure Active Directory B2C."