Amazon's Push in Education

In this interview with Campus Technology, Amazon Web Services General Manager Steve Halliwell reveals two new AWS services that will have an impact on the education market, explains why long-term data storage is becoming a challenge for researchers and shares his thoughts about the importance of "democratizing" infrastructure for students to transform education.

• By Dian Schaffhauser
• 12/10/13

Since Steve Halliwell joined Amazon as the Amazon Web Services general manager of education in 2011, he has seen engagement with universities and colleges expand from a size where he could name all of the institutional customers to a global community of 2,400 schools. "It's grown exponentially in a very short period," he said.

While researchers turned to AWS' education services early on to support their extensive compute power and storage needs and the IT organization began using cloud computing to supplement the data center resources maintained on premise, more recently, faculty and students have adopted AWS and similar offerings to gain instructional and learning resources not easily delivered on campus.

"Education is transforming as a business," Halliwell points out. "As we work with our customers in higher ed, they come to us and they say, 'I need to lower the barrier to change from an IT perspective, both financial and technical. I have to be able to change more quickly. how will I do that?'"

In this interview, which has been edited from the original transcript, Halliwell shares two new AWS services that could have an impact on the education market, explains why long-term data storage is becoming a challenge for researchers and shares his thoughts about the importance of "democratizing" infrastructure for students to transform education.

Dian Schaffhauser: The number of offerings in AWS has grown since the last time we checked in. What are some of the newer services available?

Steve Halliwell: I'd highlight two big new products that we've announced that I think will make a significant impact on the education market, either directly or obliquely in terms of what they allow people to start thinking about. The first is one called WorkSpaces, a virtualized desktop offering [that] allows folks to bring a device to their interaction with the systems and allows the IT department or administrative group to refocus all that compute that's typically out on desktops in a distributed environment into a controlled, secured, highly scalable cloud environment. WorkSpaces allows folks to move all of the heavy lifting from the client in a distributed environment into the cloud in a managed secure environment.

This is not new. Citrix and Microsoft have done this for a long time. The difference is moving it into a cloud environment and using the same tools they're using to manage their distributed environment to manage that environment within the AWS cloud. And quite frankly, the other emphasis is on cost, to be able to do these things and scale and yet lower the total cost.

That offering also plays into some of the other experimentation we're seeing in the education market, such as [massive open online courses]. Now it might possible for somebody who's designing a course that 100,000 students might take to not just design a lecture capture delivery experience with a few computer labs, but also to deliver an experience for students they can access from a multitude of devices.

One of the other new big releases we've talked about is Kinesis. Earlier this year we released a data analytics product called Redshift, which allows folks to perform petabyte-scale data warehouse services; but it's like a tenth of the cost of the traditional data warehouse. Amazon Redshift automates a lot of the common tasks — provisioning, configuring, monitoring, the backup, all the stuff researchers or administrators don't want to spend their time on, and allows them to do deep data analytics. It's what I would consider a more traditional data warehouse offering.

Kinesis is much less traditional. Think of it as a rolling health monitor or rolling dashboard. You can create a near real-time offering that allows you to be in touch with what is happening within your operation. Let's say I'm using a MOOC and I've released a class, and I decide I'm going to change the content — put a new lab out there. One of the metrics I might have in place is the amount of social interactivity that students are having. Kinesis is designed so that you put the new module out and start collecting this new information in your second, third, fourth hour of production.

Schaffhauser: When a school adopts AWS for something more traditional such as compute or storage to get through peak demands such as registration, does it have to be an all-or-nothing thing?

Halliwell: Absolutely not. That's one of the common misconceptions people have about going to the cloud. It always seems like it's an all or nothing, one-way trip. It isn't. Depending on how your application is written, you may decide to run this piece of it on the cloud and this piece of it on premise. Or you could decide that this [application] works just fine [on premise ] except a few periods when kids are coming back and registering and all you really need is the ability to flex into the cloud for the times when you to replicate data, spin up additional application servers or Web servers, and then have all those things disappear when that workload goes away.

Schaffhauser: Let's talk about the research usage of AWS. The University of California, Berkeley has a couple of research programs based around AWS, among other cloud services, including AMPLab and RAD Lab. Can you explain how researchers are using AWS?

Halliwell: The concept of AMPLab is unique in that they are applying algorithms and computer learning to a variety of different problems in the world today. They're combining machine learning and cloud computing and crowdsourcing and really pushing the envelope with regard to storing their data in the cloud and using compute resources in an auto-scale manner. So they're using intense amounts of resources only when they need it and very little when they don't need resources. They're a great use case for how researchers can benefit from moving to the cloud.

Schaffhauser: Are there other more recent programs at other institutions worth mentioning?

Halliwell: One that the [National Institutes of Health] just announced that I would say is one to watch is their Big Data to Knowledge program. That has the potential to create one of the next great computing in analytics studies out there, and will really help education come to grips with big data and other aspects of that.

We do work with those folks and are supportive of the work of NIH and [the National Science Foundation] in helping to drive innovation in research, whether it's through cloud computing or inter-institutional collaboration, or paradigm shift or sharing of data and storage of research results. We try to make sure we're as supportive as we can be there.

Schaffhauser: When a research project is over and all that data exists out there on the cloud, what happens? Where does it go? What are researchers doing with it?

Halliwell: Some researchers are choosing to keep their data in the cloud and continue to pay for it to be stored there. It's generally, less expensive for them to store it in the AWS cloud than it is for them to move it and store it in their on-premises environment, unless that's already been bought and paid for, and so they're using capacity that's readily available at almost cost to them.

The other aspect is if they put it in the cloud, they can share that information with other researchers and not have to be shipping hard drives or trying to figure out FTP for data movement. They can simply make that data available to someone else and have that person begin to use it as it works out there.

One of the challenges that we all need to look at in the long term is what [to do] about long-term research data storage. Right now if you get federal funding for your research project, you have to keep your data for X period of time after your results are published. We are very interested in that challenge and would like to be able to come to the table with a more fully formed program around storage and even the archiving of research data to meet those requirements. When you get under the covers of the research environment, when the project is over you see how much of that data does not become available to the research community in an easy, well annotated, well curated way. It can be a bit disappointing. We're certainly tuned into that and would like to help.

Schaffhauser: Let's talk about faculty use. I was interested in to see how AWS is being used for many of the data analytics education programs out there in colleges and universities. How are faculty using AWS in their courses?

Halliwell: Let's take a look at an educator. Perhaps somebody is running a class and has an interest in leveraging the cloud. Maybe it's computer science, data visualization, something in a graduate school of business. We may see 30 students at X school have just signed up for an AWS accounts. They go to the AWS page, sign in, put their information in, and get an AWS account and use that AWS account for the class. Or the professor can administrate those accounts, set the accounts up for the students. It all depends on the content being taught. For instance, if you're teaching a class on cyber security, it might be better to have the student accounts be separate and have them run their analysis or their intrusion exercises in separate accounts. However, if you're teaching a course and it's collaborative in nature, you could use one account and have 30 different students interact within that account with our role provisions; you can spin up as many compute nodes on the back end as you need to support the 30-student interaction. On the first day of class, people sign up and get their AWS account and get access to the resource, whether compute or storage, and they're off and running and interacting.

We are at work on an enhanced coursework program, where we hope to better support professors teaching classes [involving] AWS, and it's something we hope to announce in the coming months.

Schaffhauser: Do you mean providing curriculum that they could adapt for their own courses?

Halliwell: I think there will be some of that. Some of it may be allowing them to search for content in an easier fashion or share observations. And maybe [there will be] an online community for professors that are using the cloud for teaching, so they can share best practices.

Schaffhauser: Amazon Web Services also has a program that lets student organizations or students doing special projects set up accounts for those. How are students using that?

Halliwell: We see students using that program to complete part of their formal curricula: "We need some compute and storage to do our class projects, and we decide we're going to use this data in AWS and spin that up and do that." Or it can be directed by the school or professor. There's also less formal use of that, where students may not actually have an assignment, but they have something they want to develop, something they want to create. Like-minded individuals decide they want to go build something or analyze something and see what happens. That is what our student program is designed to support.

Most importantly, it allows the student that may not have huge funding, may not have natural access to supercomputers or the best on campus lab environment to have the same amount of access as the compute and storage resources found in some of the most elite schools and research universities in the world. They've got the full power of all of Amazon Web Services data centers at their disposal.

I look at that as an opportunity to help democratize the use of technology in education. Schools that haven't historically invested in computer science or big on-premise labs can offer an educational experience to those students that rival some of these very elite schools that have everything you could possibly want in terms of technology.

Schaffhauser: The AWS grant program is dazzling in its simplicity. Talk about that for researchers, faculty, and students.

Halliwell: We have tried to keep the grants program as open as possible, meaning that the form basically has an open text field and someone can fill that in and tell us what they'd like to use AWS for, and we will assess that request. We look at those submissions as they come in and we learn a lot about how people are using AWS. But we also ask, as they use the program, to give us feedback on how it went, what kinds of results they achieved. We love to see their published work and see what they were able to do and continue to support research in a broad way.

Schaffhauser: As a percentage, how many grant applications do you approve?

Halliwell: I'm not at liberty to say, but I can tell you, we are very supportive of the applications that we receive.

Schaffhauser: Amazon Web Services talks about how simple it is to get these cloud services going, but the company also promotes education solution providers. What are those companies doing that I wouldn't be able to do myself?

Halliwell: I can answer that in two ways. Let's go through a scenario where a school wants to begin using a cloud. Schools are big enterprises. They have all of the challenges that exist in enterprises in the business world or in government with regards to their funding, their reporting requirements, their security paradigm, billing. Some of these companies are truly expert at advising their customers on how to structure and best use AWS resources. These partners will help you develop a workflow and approval process combined with accounts structure all the way through to billing and reporting and tying back to chargeback or showbacks or grant reporting, and those types of requirements.

When schools are used to a classic IT paradigm, where they buy hardware and it's a capex cost and the cost doesn't change appreciably whether the machine is used 5 percent or 100 percent, there are processes and disciplines in place to work there. When you move to the cloud, there are some new muscles you need to exercise. For instance, if you're using five percent of compute resource or using 50 times that in the cloud, your bill can go up or down. So learning how to manage a utility-based environment is a really valuable skill — something that enterprises like universities and K-12 need to be able to internalize. Sometimes schools will turn to industry experts in the form of these partners to advise them on the best way to do that.

The second way that education entities benefit from working with these partners is that many of them deliver a solution that runs on top of AWS. They have expertise within that area that AWS does not have. We're creating those building blocks, on compute, storage, networking, and security. But they're experts in learning management systems or student information systems. and AWS doesn't do that type of work. There are some great companies out there that have true depth of experience within their solution that can add real value for schools around the world.

I do think we'll be doing a lot more work in allowing students that don't have access to on-premise resources the ability to connect to a supercomputer from any school on the planet. That's a key part of what cloud provides to education that has been overlooked. Democratizing access to resources in education and allowing the brightest minds from small schools or community colleges or leap schools to do the best work they can is something we're excited about and that we think can help transform education as well.