Boosting Microsoft’s transaction platform by migrating to Microsoft Azure

Oct 27, 2021   |  

Microsoft Digital technical storiesMigrating to Microsoft Azure was just what Microsoft needed to transform its massive purchasing system.

The Commerce Transaction Platform (CTP) that the company depended on heavily for more than 20 years was a powerhouse, but it was aging—the Commerce and Ecosystem (C&E) team that managed it needed to move it to the cloud to get the boost in performance and reliability that they were looking for.

If the system goes offline, loss is measured in thousands of dollars per minute. It has to be certain, it has to give you an exact amount, and it has to be available to reliably process all the transaction requests.

—Ivan Trindev, principal group engineering manager, Commerce and Ecosystem team

Migrating to Microsoft Azure would have another benefit as well.

“There’s a lot of flexibility with Microsoft Azure,” says Ivan Trindev, a principal group engineering manager with C&E. “That would mean we could run our business comfortably with a lower cost of maintenance.”

To get specific, the CTP manages escrow of purchases through Microsoft’s online stores and services, including Microsoft Office 365, BingAds, Windows Store, Microsoft Dynamics 365, Microsoft Azure, and more. When a customer goes to complete a transaction, the CTP makes sure Microsoft has collected the money as reported revenue and verifies that the sale is final. The platform also handles billing for user subscriptions.

“If the system goes offline, loss is measured in thousands of dollars per minute,” Trindev says. “It has to be certain, it has to give you an exact amount, and it has to be available to reliably process all the transaction requests.”

Migrating to Microsoft Azure would mean that the CTP would have the performance to handle the growing number of transactions and the reliability to give C&E peace of mind.

The move didn’t happen overnight.

[Learn how Microsoft used Azure to retire hundreds of physical branch-office servers. Discover how Microsoft migrated critical financial systems to Microsoft Azure.]

Configuring the platform for success

With more than six petabytes of data and 700-plus machines, the CTP was an on-premises force.

This vast array of SAN storage networks and processors required a refresh every few years to keep up with growing system demand. Migrating to Microsoft Azure meant C&E’s ritual of evaluating system needs and projecting for the future could be an exercise in digital transformation.

Migrating to Microsoft Azure would allow them to keep all that power and be able to do things like pivot in a new direction or upgrade part of the service much more easily.

Four different functional components make up the CTP: online systems, backend processes, data storage, and revenue recognition. Each requires different degrees of computing power, by using a Microsoft Azure Dedicated Host (ADH) and matching existing systems to the appropriate virtual machines (VMs), C&E was able to reproduce the infrastructure they needed for the transaction platform.

“We didn’t have to downsize our needs,” Trindev says. “Azure matched them.”

Some of the legacy systems needed to run the CTP led to C&E selecting an Infrastructure-as-a-Service (IaaS) environment to build from. With several options to choose from, C&E decided on Microsoft Azure SQL—Virtual Machines to handle their huge processing needs and scaling requirements.

“Comparing on-premises to Azure, we realized we’d need the largest option available to us on IaaS,” says Pradeep Byreddy, a principal service engineering manager with C&E. “This was exceeding the budget allocated for CTP and had to work with Azure in building a cost-effective SKU which still accommodates the scale and reliability requirements we have.”

Having selected the largest computing option to handle some of the more demanding tasks, C&E could rely on virtual machines with less capacity to manage the rest. Similarly, while the fast but more expensive ultra-disks were closer to the SAN storage solutions C&E had previously relied on in their on-premises environment, they found that Microsoft Azure’s premium SSDs met their performance benchmarks without adding an extra financial burden to the platform.

Having defined the Microsoft Azure infrastructure needed to host the CTP, C&E could now move on to shifting the six petabytes of data from on-premises to the cloud.

Navigating the hybrid experience

Where a new business starting fresh would never encounter the delicate balance of managing simultaneous on-premises and cloud environments, migrating to Microsoft Azure required C&E to carefully orchestrate a hybrid strategy.

This approach required the CTP to exist in Microsoft Azure and the legacy on-premises environment at the same time, with the systems synchronized for testing purposes. To do this, C&E relied on SQL to create a seamless transition between the old and the new. Backups of the on-premises environments were restored in Microsoft Azure to the ADH VMs, then synchronized.

“This enabled a seamless failover approach,” Byreddy says. “We didn’t see any issues; it was just like any other maintenance failover.”

C&E carefully moved one component at a time using ARM templates. With the side-by-side approach they could test customer experiences and data outputs generated in the cloud. Finding values and experiences to be the same, C&E knew they could count on Microsoft Azure to report transactions reliably.

Reaping the benefits of Microsoft Azure

C&E needed high availability, performance, and reliability. They got that and more by migrating to Microsoft Azure.

Optimizations across Microsoft Azure allow C&E to scale services up and down based on what the CTP needs. Unlike the on-premises environment, C&E can procure and decommission virtual machines based on system demand, giving them a cost-efficient way to keep the CTP running. This cost-efficient way of running the CTP is even more pronounced as on-premises servers and hardware become more expensive.

Monitoring is another advantage of Microsoft Azure. We’re also using platform features to introduce self-healing, which helps reduce the downtime of any incident.

—Pradeep Byreddy, principal service engineering manager, Commerce and Ecosystem team

C&E can also upgrade and add new features on the fly.

“We didn’t have flexibility before,” Trindev says. “Change management and procurement took months. Now we can be dynamic.”

The team no longer worries about maintenance costs or fixing broken hardware; that falls on the Microsoft Azure team. And as the C&E team spends more time in Microsoft Azure, they’re finding new processes and practices to introduce additional savings, including utilizing native systems that reduce reliance on third-party services.

“Monitoring is another advantage of Microsoft Azure,” Byreddy says. “We’re also using platform features to introduce self-healing, which helps reduce the downtime of any incident.”

Looking ahead to the CTP

Now that C&E has moved Microsoft’s transaction platform to the cloud, they have an environment that’s more reliable than their on-premises counterpart, performs better, and can scale in a cost-efficient way.

The team has relieved itself of having to spend significant time and resources responding to the on-premises refresh cycle.

“There was zero negative impact in moving from one to the other,” Trindev says. “Now, when I think about upgrading a component, it’ll take two weeks. It used to take months.”

Product features in SQL and Microsoft Azure work right out of the box, allowing the team to leverage new features to improve the CTP. Eventually, C&E will shift from IaaS to a Platform-as-a-Service (PaaS). This move will introduce even more flexibility and ease of management.

“This is the value of having partners across Microsoft,” Byreddy says. “SQL will enable a seamless migration to PaaS.”

Key Takeaways

  • Every refresh cycle is an opportunity to review infrastructure needs. Should you stay on-premises or is it time to migrate to Microsoft Azure?
  • Compare your current performance and storage utilization to a variety of virtual machine benchmarks. Sometimes it won’t be a one-to-one equivalency, but testing will reveal your capacity requirements.
  • PaaS is more flexible and easier to manage than IaaS, but sometimes migrating to Microsoft Azure means starting with IaaS as a transitory step towards PaaS.
  • Premium services within Microsoft Azure might seem expensive but doing a careful analysis of the cloud’s reduced overall maintenance and servicing costs will reveal offsets.
  • Once you’re in Microsoft Azure you have a lot more flexibility than on-premises. In addition to being able to scale up and down virtual machines based on need, it’s easy to deploy new services and features.

Related links

Tags: ,