Distributed Computing Systems Group

Tiera: Towards Flexible Multi-Tiered Cloud Storage Instances


Many cloud providers offer a range of storage services, each of these offers different tradeoffs between cost, performance, and durability. Applications typically need to combine different storage services (local or edge storage with Amazon S3 or Windows Azure for example) to realize a tradeoff suitable for their requirements. This task of combining different storage to realize composite benefits is tedious, requires experimentation with the different storage options and writing complex code in each application to glue together the different storage options.


Tiera is a system that enables an application to specify flexible storage instances that comprise of different kinds of storage (eg.Amazon S3, Windows Azure, Edge Storage etc.) and a dynamic data storage and placement policy to achieve the desired tradeoff between cost, durability, and performance.



Research Issues

  • Scalability, handling millions of requests and being able to store billions of objects
  • Translating user supplied policy to an optimized static binary
  • Client workload analysis and prediction
  • Modifying policies at runtime to adapt to changing client workload
  • Achieving fine grained QoS guarantees in a multi-tiered cloud storage system, where individual storage might violate SLAs due to poor performance isolation


  • Tiera: Towards Flexible Multi-tiered Storage Instances Poster
    Presented at the "Technology Forum and Open House Event" organized by
    Department of Computer Science and Engineering,
    University of Minnesota, October 2013.