Social media platforms and dozens of other web and mobile apps generate countless amounts of data. And with a constant flux of data comes a continuous need for high-performance and highly scalable ways to store and access this data. Yue Cheng, an assistant professor in the Department of Computer Science, won a National Science Foundation CAREER Award to meet the needs of the big data industry.
Cheng and his research team will use the nearly $580,000 award “Harnessing Serverless Functions to Build Highly Elastic Cloud Storage Infrastructure” to develop a scalable and cost-effective cloud computing storage system using serverless computing.
“We are researching new ways of using emerging serverless computing capabilities to build a high-performance cloud storage infrastructure,” says Cheng. “Serverless computing is the next generation of cloud.”
Cheng’s new infrastructure, InfiniStore, will reduce the cost and need for manual storage management, says Cheng.
Conventional cloud computing uses virtual machines (VM) that are rented out to cloud users. “The user pays for whatever resources they need, and while it gives great flexibility, the VM-based model isn’t the most effective nor elastic. Cloud service providers still charge for space and capacity that is reserved but not in use. Cloud users also must manually start or stop virtual machines to best suit their needs.”
Serverless computing is the solution to cost and resource issues under the conventional VM model. Because of its inherent elasticity, it can scale up and down autonomously. Since it doesn’t require a space reservation, like with VMs, Cheng can implement a new cloud storage pricing model based on usage.
“Let’s say at one point in time there are no requests, and no cloud function is launched. And a later point in time, there is a huge spike in data access requests. The elastic storage would automatically trigger thousands of cloud functions, each serving as a tiny little data storage unit, to serve this spike of data access requests, and the cloud user is only charged when the data stored in cloud functions are accessed,” says Cheng.
Cheng’s project includes collaborations with industry leaders like NetApp and IBM Research and fellow researchers at Mason. In addition to the development of InfiniStore, the grant also consists of an educational plan centered around teaching serverless computing and increasing access for undergraduate students to participate in leading research.
“We will design a new cloud service called InfiniCloud, which is a digital and interactive notebook service that allows students and educators to implement, write, and deploy serial and parallel Python programs at any scale,” says Cheng. “We will also start an outreach plan to promote diversity in computing and engage undergraduate students in computing research.”
Cheng stresses the importance of advancing computing capabilities and allowing students and educators access too. “This project will support big data storage, which is extremely important, but I am also committed to introducing undergrad students to this state-of-the-art field in cloud computing,” says Cheng.
“Yue’s NSF CAREER Award comes at a time of nearly universal reliance on cloud services to support critical enterprise functions. His research will realize significant improvements in the ways that cloud resources are allocated to end-users, through innovations in cloud storage technology,” says Department of Computer Science Chair David Rosenblum. “And the educational plan he has devised will give Mason students significant hands-on learning opportunities with this important area of computing technology.”