New Google service aims to move mainframe workloads to cloud

Mainframes and the cloud

Dual Run, available in preview globally, enables parallel processing of workloads in a customer's on-premises mainframe and within GCP at the same time to simplify the migration of mainframe applications to the cloud.

The new service builds on cloud migration tools developed by Banco Santander, an international bank headquartered in Spain. The bank moved workloads previously isolated to its mainframes into Google Cloud.

The service enables customers to test and optimize cloud workloads without affecting primary mainframe computing before moving these applications into Google Cloud as the system of record, according to Google. The service also enables customers to keep existing mainframe infrastructure available as a backup or fallback system after a migration.

Google said Dual Run will eventually include migration strategies and options for customers in banking, healthcare and the public sector, industries that still rely on mainframes for workload processing and data residency requirements.

Cloud hyperscalers such as Google pitch compatibility and interoperability across cloud and on-premises infrastructure, Raffo said. What makes mainframes different, however, is how many are intricately tied into the levels of workload and application performance an enterprise may demand, as well as regulations regarding data residency.

Moving workloads to the cloud will require an attractive price point with a minimal impact to performance before any enterprise might consider the headache of migration, Raffo said.

Service providers such as Accenture and Kyndryl will support Dual Run implementations, according to Google.

Faster cloud compute

Google also unveiled the new C3 machine series, growing the hyperscaler's portfolio of VMs.

The C3 machine series, available in private preview, adds new performance capabilities to the hyperscaler's Compute Engine lineup, which targets compute-intensive workloads for high-performance computing. Existing compute-intensive products include C2 and C2D, which run VMs on Intel or AMD CPUs respectively.

The new product uses the fourth-generation Intel Xeon Scalable processor and Google's Intel Infrastructure Processing Unit, which Google said has provided early adopters a 20% increase in workload performance compared with the C2 VMs.

The C3 machine series follows the announcement last month of Google Cloud Hyperdisk, a new block storage service.

Sid Nag, an analyst at Gartner, said announcements such as C3 and the other catalog updates from Google provide incremental improvements and support Google Cloud's focus on DevOps and applications. It's a strategic differentiation compared with market-dominating AWS and Azure.

"Google is pretty good at technology as a whole," Nag said. "They're showing growth in the revenue as a whole, [but] they're still third in the market."

Data center expansion

Google is also expanding its regional data center footprint to five new countries: Austria, Greece, Norway, South Africa and Sweden.

The company previously opened five other regions this year in Columbus, Ohio; Dallas; Paris, France; Milan, Italy; and Madrid, Spain. Additional planned regions for expansion include Malaysia, Mexico, New Zealand and Thailand.

Google said the expansion is necessary to meet growing demand and brings the global network to 48 live and pending regions.

Tim McCarthy is a journalist living on the North Shore of Massachusetts. He covers cloud and data storage news.