Storage systems keeping data safe

Building blocks of modern life

Storage systems are the fundamental computing building block of our modern lives – and like physical building blocks, computer storage systems must function well in fair weather as well as sudden disturbances to keep data safe.

“Storage robustness is crucial to systems ranging from financial institutions, where downtime can result in millions of dollars lost, to scientific computing, where the generated data advances any number of challenges facing humanity,” said Zheng.

The problem of system robustness and crash consistency is decades old, but creating realistic test crash events isn’t easy and the task grows ever more difficult as storage technology becomes more complex. Zheng will address the longstanding problem by creating a new flexible, scalable framework for thoroughly testing the crash consistency of many different storage systems.

Measure, model, make automatic

Zheng plans a three-part approach. He’ll design a comprehensive suite of benchmarks that can be used to drive storage stacks to vulnerable states and measure crash consistency. Next, he’ll leverage virtualization technology to model and generate realistic crash events for triggering vulnerabilities.

Zheng’s team will also create a crash consistency exploration engine that will enable automatic testing. And all work will be done in partnership with leading industry storage system makers to ensure tools are useful to real-world challenges.

“My belief is that storage systems should never lose data, no matter what happens. Combined together, these developments and collaborations are a crucial step toward the goal of truly robust storage systems,” said Zheng.

NSF CAREER Award Winner

Mai Zheng, assistant professor of electrical and computer engineering