Mathias Rosenthal - Fotolia

AWS ERP migration required staff buy-in and curiosity

East Tennessee State University is moving its ERP systems from its data center to Amazon Web Services. It's a big project, and essential to its success was getting the IT staff on board.

East Tennessee State University is undergoing an AWS ERP migration, a major change for the university's IT department. One of the hardest parts of the move was getting the IT staff to warm to the idea.

In this Q&A, Andrea Di Fabio, the university's chief information security officer and associate CIO, discusses the ongoing Amazon Web Services (AWS) ERP migration with SearchERP. The university is moving an on-premises system that runs Banner by Ellucian ERP to AWS. This is otherwise known as an infrastructure-as-a-service migration. Banner is commonly used by colleges and universities.

The university, which has about 3,000 employees and 15,500 students, uses its ERP system to manage the student information systems, as well as the financial system, procurement and HR. They are using Docker containers and Terraform to design their infrastructure.

Andrea Di Fabio, associate CIO & chief information security officer, East Tennessee State UniversityAndrea Di Fabio

The cost of running the school's ERP on AWS, as compared to keeping it on premises in university-operated data centers, will be roughly similar over five years. But the university will gain improvements in security and disaster recovery and free up IT staff time, now spent on server physical maintenance, for new projects, Di Fabio said.

Di Fabio recently gave a presentation on the project at a CompTIA conference. In a separate interview, he explains some aspects of the ERP migration project. What follows was edited for length and clarity.

What are you trying to fix with this AWS ERP migration?

Andrea Di Fabio: I have two hats. One is to drive risk compliance and security at the university. The other one is to drive our technical team to think outside the box to improve technology and improve processes. The university had two data centers, a primary and a secondary. It had just built a third data center, which was becoming the primary data center. We had a lot of legacy servers and needed to buy new ones. I started asking questions and challenging my technical staff. 'What if we don't buy anything? What if we move to the cloud?' That sparked their (the technical staff's) interest. 'What could happen if we do this?'

What happened once you started asking questions?

Di Fabio: When you start asking questions, the status quo comes out: 'We've been doing this. We feel comfortable with this. It's working. Why? Why move? Why go through the trouble of learning what the cloud is all about and then rebuild our infrastructure in a place that is unfamiliar to us? Why would we move to someone else's data center?' I was just asking open-ended questions. Our CIO started asking questions about the budget: 'What does this mean to our infrastructure, to our cost model?' So, that's when the project became a little bit more serious. It went from 'Let's ask questions' to 'Let's start planning for the AWS ERP.'

What are you doing?

Di Fabio: We are going to an AWS ERP infrastructure-as-a-service model. We're taking a computer that is in our data center and moving it to another data center. That's the 10,000-foot view. To do it, we completely re-engineered the infrastructure of our system by taking advantage of a cloud-based infrastructure. We used Docker containers for many processes. A container is almost an operating system built just around that application. It's stripped down to its bare bones so they can just run that thing. One, it is more secure. Two, you're adding modularity to it. Three, it separates the application from another application. We are going to analyze each and every application that comprises this humongous ERP system. What is the cloud way to run that application? It's a different set of mind. Can we rebuild the infrastructure in a way that's more modular? We had no modularity in our data center. We were slapping applications on servers.

How did you design this?

Di Fabio: We used Terraform, which is a cloud infrastructure orchestration platform. Think of it as a file that specifies how to build your data center. The infrastructure now becomes code, which means I can review the whole infrastructure in front of me. It can be audited. Infrastructure that's repeatable -- infrastructure that everyone can look at. Anyone can look at the code if they understand how Terraform works. The configuration management system we use is Puppet.

You had trouble in selling this idea. How did you overcome it?

Di Fabio: There was a lot of skepticism in the beginning. I asked the CIO for a packet of money, and I told our technical guys to just start playing with the cloud. There is nothing better you can do to technical guys then to give them money to play with technology. All of a sudden, they came back and started saying, 'You know what? This makes a lot of sense. There is a lot of pieces that we can move right now to the cloud in a very simple manner.' I went back to the CIO and said, 'I think our team is actually excited about this. They're ready for this.' We started testing with our testing and development environment and one little piece at a time. They started moving the development environment one piece at a time. That's how it happened.

What are the lessons learned from your ERP migration?

Di Fabio: In terms of lessons learned, our procurement office was not ready to switch from Capex to Opex. They're not set up that way. Earlier discussion with procurement is something that organizations definitely need to look at. The other main lesson learned is the legal implication. Early discussion with the legal department is a must before you start looking at sensitive information moving to the cloud. [The university ended up making an agreement with a company that resells AWS that also has expertise in meeting the legal obligations for moving sensitive data to the cloud.] The procurement and legal issues are definitely a lot more complicated and time-consuming than the technical issues. You give a technical person that is excited about the technology access to the technology, they will figure it out pretty quickly. They will train. They will self-train.

Dig Deeper on ERP administration and management

Data Management
Business Analytics
Content Management