Nomad_Soul - Fotolia

News Stay informed about the latest enterprise technology news and product updates.

BofA confronts OpenStack problems in data center overhaul

Bank of America has solved enough of OpenStack's troubles to start running production workloads on the platform next year.

Bank of America Corp.'s massive shift to a software-defined data center (SDDC) encompasses two architectures: one built by a vendor and the other by an open source community. So far, the vendor is winning.

Very few of the 1,500 workloads Bank of America (BofA) is moving to the new architectures every month are running on OpenStack, an infrastructure as a service platform freely available under a license from The Apache Software Foundation. Instead, the bank has bypassed OpenStack problems by running the 6,500 workloads in production on proprietary SDDC technology from a vendor it declined to name.

"We found earlier this year that the OpenStack piece wasn't ready, even for the [noncritical] workloads we're running now," said David Reilly, the global technology infrastructure executive at BofA, based in Charlotte, N.C.

As the nation's second largest bank, with $2.1 trillion in assets, BofA can afford to tackle cutting-edge technology, becoming a bellwether for enterprises with fewer resources. BofA's goal is to virtualize all elements of its infrastructure -- networking, storage, computing and security -- so the bank can quickly modify applications or deploy new ones, as customer demands change.

Having a high level of flexibility is key to how BofA operates today. Customers are less interested in doing business in the bank's branch offices. Instead, they want to deposit and move funds using a PC or mobile device, and withdraw cash from ATMs anytime of the day or night. Providing customers with those services, and adding new ones, requires a highly adaptive data center.

Solving OpenStack problems

Reilly believes his army of engineers has solved the major problems with OpenStack. Starting early next year, the bank will move production workloads to the platform. How many of BofA's roughly 70,000 workloads will end up on OpenStack or the proprietary technology is not known.

"There is a place for both driven by workload type and criticality," Reilly said.

Today, the bank runs mostly "low impact and low priority" workloads, such as those related to software development and testing, on the new infrastructure, Reilly said. Next year, BofA plans to migrate more important workloads from management systems related to IT and customer relations.

OpenStack's problems were not connected to its core technology, which worked "very, very well," Reilly said. What the platform couldn't do, and the bank had to fix, was run thousands of workloads.

OpenStack had to run those workloads while letting the bank's other systems maintain constantly changing records of all the institutions' applications. Records include details on related servers and storage. Without that information, troubleshooting problems would be nearly impossible.

It's tough to compete with free.
David Reillyglobal technology infrastructure executive at BofA

The bank also had to build management tools that spanned the proprietary and OpenStack platforms. The system, which is scheduled for deployment this month, automates the process of deploying a workload on either the proprietary environment or OpenStack, depending on which is more suitable.

"If we're going to migrate 1,500 workloads a month, we need this to be done without a human hand having to touch it," Reilly said.

The management system also lets developers provision new Windows or Linux virtual machines (VMs), or re-provision those that have been adjusted to changes in applications. The tools also help in choosing storage and networking elements.

"Getting as many hands out of the provisioning process -- the management process -- and having that done through automation is super critical," Reilly said.

BofA's commitment to OpenStack

Reilly remains committed to OpenStack, while acknowledging the need for proprietary systems to run alongside it. Proprietary technology is better suited for stateful applications and those that are storage-intensive.

"That might well change, but no, I'm not on a march here to displace the third-party platform," he said. "I think it has a place in our environment."

As OpenStack matures, the bank will move workloads from proprietary systems to the open platform, if it is up to the task. "It's tough to compete with free," Reilly said. "The price differential is going to start to play a part, as the capability gap narrows."

So, while proprietary software has the advantage today, free is likely to make OpenStack a winner tomorrow.

Next Steps

Inner workings of the 'Swift' OpenStack Object Storage

An OpenStack guide for cloud providers

Deciding whether OpenStack is right for an IaaS cloud

Dig Deeper on Software-defined data center

PRO+

Content

Find more PRO+ content and other member only offers, here.

Join the conversation

3 comments

Send me notifications when other members comment.

By submitting you agree to receive email from TechTarget and its partners. If you reside outside of the United States, you consent to having your personal data transferred to and processed in the United States. Privacy

Please create a username to comment.

What are the difficulties your organization confronted in using OpenStack?
Cancel
Is BoA using the trunk version of OpenStack, or one of the vendor distributions (e.g. Red Hat or Mirantis)? Given how some of the large companies have seen their OpenStack teams leave to other jobs, it'll be interesting to see how they keep this team in place. 

OpenStack definitely needs some Enterprise customers to be willing to speak about their Private Cloud environments.
Cancel
Interesting point and something worth watching. I believe BofA is using the trunk version of OpenStack. I'm sure the bank is paying really well to keep talent.
Cancel

-ADS BY GOOGLE

SearchNetworking

SearchEnterpriseWAN

SearchCloudProvider

SearchUnifiedCommunications

SearchSecurity

  • Passive Python Network Mapping

    In this excerpt from chapter two of Passive Python Network Mapping, author Chet Hosmer discusses securing your devices against ...

  • Protecting Patient Information

    In this excerpt from chapter two of Protecting Patient Information, author Paul Cerrato discusses the consequences of data ...

  • Mobile Security and Privacy

    In this excerpt from chapter 11 of Mobile Security and Privacy, authors Raymond Choo and Man Ho Au discuss privacy and anonymity ...

SearchDataCenter

Close