How Notre Dame is going all in with Amazon’s cloud
- 14 December, 2015 19:41
Higher education websites can be suddenly and unexpectedly thrown into the national spotlight. Think of a college football team winning a championship or a natural disaster striking a campus. In either situation, a related traffic spike can flood a website.
These were some of scenarios that leaders of Notre Dame’s Office of Information Technology (OIT) thought about three and a half years ago when they load tested the university’s website, which had been hosted on a couple of servers at the main campus and backed up at a nearby collocation facility. The site failed its test.
“In times of crisis, you want to have confidence that your emergency communication vehicles can hold up if they’re basically getting DDOS’d (distributed denial of service) with people looking for information,” says Sharif Nijim, an enterprise architect in Notre Dame’s OIT. “We needed to do something.”
OIT officials explored partnering with other universities, or expanding their collocation footprint. Nijim and his coworkers also reached out to Amazon Web Services to explore whether the company’s Infrastructure-as-a-Service (IaaS) cloud could be a fit.
For years prior, the university had used Software-as-a-Service (SaaS) tools. It was an early adopter of Google Apps, first as an email platform for students and later staff. It uses Box for productivity, and Concur for travel and expense management.
During an introductory demonstration with AWS, company officials spun up an entire CRM system with multiple databases in less than an hour, then spun it right back down. Nijim was sold, and later other OIT officials agreed to migrate Notre Dame’s website and mobile app into AWS.
Soon after its re-launch in 2012, the site got its first test: Notre Dame in early 2013 played in the BCS National Championship football game. Despite the Fighting Irish losing to the University of Alabama, the website stood strong in the face of massive demand.
“It performed exactly as it was designed to,” Nijim says. The website automatically got virtual machine instances and load balancers when demand spiked, then spun them down after the stress period ended. “That made believers out of a lot of people who had been skeptical.”
From there, Notre Dame was all in on the cloud, and its OIT has since adopted a cloud-first policy.
“It’s really one of our core IT strategies,” says Michael Chapple, senior director of IT Service Delivery at the university. “It’s the biggest technology change we’ve made in the last 10 years. We’re reinventing the way we provide IT service to the campus.”
Cloud-first doesn’t mean cloud-always though. When a need arises, OIT officials first look for an existing SaaS solution. If there isn’t one, then they turn to the public IaaS cloud. And when IaaS isn’t an option, Notre Dame will look to host workloads and applications on campus.
A journey to the cloud
Deciding to go to the cloud and actually doing it are two different things. The move has required a big transformation within the roughly 240-member OIT team. “It takes a major mind shift,” says Chapple, who teaches a course on Information Security and is a former National Security Administration researcher.
Prior to adopting cloud, Notre Dame’s OIT included a team focused on security. That team heavily researched AWS and worked with Amazon architects to secure the company’s cloud-based systems. One key, Chapple says, is that Notre Dame takes advantage of automatic reporting capabilities AWS offers. But OIT workers also built a customized tool that analyzes AWS usage and issues alerts when any abnormal activity is detected or if a configuration doesn’t match a pre-determined best practice. (AWS has recently rolled out its own tools named Inspector and CloudWatch that provide some of this functionality.)
Another key to success has been setting up a team of about 25 members within OIT to work in a dedicated area named Cloud Central. The team is focused exclusively on cloud migrations and implementations, and includes workers from various other groups – for example, infrastructure and operations professionals, system admins, security experts and storage staff. Notre Dame has heavily trained these workers by sending them to conferences, arranging briefings by AWS officials and having them be the main users of AWS’s cloud at the campus. It has become the go-to team for all things cloud and AWS at Notre Dame.
Setting up either an internal center of excellence or hiring a third-party consultancy to act as a central point of contact for cloud usage is a direction other organizations have taken as well. Retailer Nordstrom took a similar approach for its use of AWS.
Michael Chapple, senior director of IT Service Delivery at the University of Notre Dame
Notre Dame constantly rotates members of the IT staff through Cloud Central so they can get a feel for what cloud-first team operations are like.
“We’re keeping localities of expertise around specific disciplines,” Nijim says. There’s a platform group responsible for configuring the operating systems being used and ensuring they’re correctly patched. Most workloads in AWS are spread across multiple Availability Zones, but certain ones are replicated across more than one region to ensure resiliency – and there’s a Notre Dame OIT team that manages that.
Notre Dame’s OIT hasn’t changed size since the school adopted the cloud, but roles within the organization have transformed. Storage operators who used to manage infrastructure racks now work to ensure proper use of cloud-based storage options. One employee who used to manage the data center operations is now a point of contact for monitoring AWS’s usage trends, optimizing virtual machine instance sizes, and ensuring AWS resources are spun down when no longer being used.
Lessons to be learned
Notre Dame isn’t unique in its heavy usage of public cloud services, but it is distinctive in the higher education world.
“They’re the first nationally recognized university to really come out full force with a whole set of schematic diagrams and processes and controls in place to say we are a cloud first organization,” said Tom Dugas, associate director of computing services at Carnegie Mellon University, who is also co-lead of the Educause Higher Education IT Organization's Cloud Computing Constituent group, a working group for academic IT workers (see more from Dugas in: “How Higher Ed IT is Staying Ahead of the Cloud Curve”).
Organizations outside of academia have also begun wholesale migrations to the cloud: Netflix is a poster-child of heavy AWS usage, and startups like AirBNB, Pinterest and Slack are hosted in Amazon’s cloud. More and more enterprises are open to the idea, including publisher Conde Nast.
How can other organizations do what Notre Dame did? Chapple says it’s key to start small and grow. Notre Dame officials didn’t wake up and one day decide to go cloud-first. “Find a key service where it makes clear business and technology sense to use the cloud and do that first,” he says, harkening back to Notre Dame taking its website and mobile app to the cloud.
Workloads that have varying infrastructure requirements that scale up and down a lot are ideal for cloud hosting. So are workloads that are isolated from other systems. During the next two years, Notre Dame plans to move more sensitive and secure workloads to the cloud.
That doesn’t mean everything’s going to the cloud though. Chapple says the university’s on-campus telephone system will likely stay put – he doesn't want that to be dependent upon an outside connection to access it.
Many organizations value IaaS cloud computing because of the agility it can bring: Resources can be scaled up when needed and spun down when not. This is helpful during the peak of class registration, for example. Nijim says: “We’ve turned into a reliable service provider for the campus, and that’s 100% because of how we’ve built our platform in [the cloud].”
Network World’s Bob Brown contributed to this article