Lorenzo Alvisi, Byzantine Replication for Trustworthy Systems.
This project will investigate how to improve the trustworthiness of networked
information systems by using redundancy to simultaneously enhance integrity,
availability, and confidentiality despite malicious attacks that
exploit software bugs. We revisit the state-machine approach, the most
general technique for building replicated services, and suggest new
solutions that 1) reduce by a third the degree of replication required to
tolerate a given number of compromised replicas and 2) show, how,
counterintuitively, replication can be used to help, rather than harm,
confidentiality.
If successful, this research will develop techniques that can be used to
protect the privacy of citizens, the operation of businesses, and the safety
of national infrastructure as networked information systems become increasingly
central to mission-critical and business-critical systems.
Mike Dahlin,
Dynamic Replication for Robust Cyberinfrastructure Services
Our goal is to support robust cyberinfrastructure services that are as simple to deploy and maintain as centralized
web services but whose availability and performance are much better than traditional services. If successful, this
research may have far-reaching effects by changing the way that a broad range of applications and services are
constructed, deployed, maintained, and used.
We hope to meet this goal by developing fundamental and broadly applicable techniques to improve the
dependability of large-scale distributed services. In particular, our research vision is to enable self-managing
systems where a service can be installed and maintained at one location while it "flows" to different locations on a
grid of resources in response to user demands, quality of service goals, network conditions, network and server
failures, and resource availability. Our initial work suggests that a key challenge to achieving this vision is data
replication: service replicas must operate on a common but distributed collection of data while still providing a
consistent view of the replicated service.
In order to address this challenge, we will develop a robust, self-tuning data replication framework that extends the
state of the art by improving consistency trade-offs compared to existing systems, by making better use of cheap
bandwidth and disk storage, and by dynamically replicating data to where replicas can improve system performance.
Harrick Vin,
An Adaptive Run-time Environment for Network Systems.
This project will investigate techniques for building highly scalable
server clusters capable of co-hosting efficiently a large number of
services simultaneously. There are two aspects to this problem. First,
cluster resources must be allocated to services based on their current
demand. Second, the load across servers must be balanced such that the
performance of each service scales linearly with the cluster resources
allocated to the service. We address both problems.
The outcome of this research will be new algorithms, architectures,
and prototype implementations of highly scalable server clusters.
