Microsoft Research

Decentralized approaches spanning multiple administrative domains (MAD) are an increasingly effective way to deploy services.

Current generation Web applications run at scales of hundreds of millions of users, each with thousands of objects, requiring petabytes of data storage, even excluding large objects such as images

Linearizability: The behavior of concurrent objects is best described through their safety and liveness properties, often referred to as correctness and progress.

In this lecture, we cover the fundamentals of distributed message-passing algorithms with an emphasis on their correctness.

ACID transactions are a classical tool that makes it easier to develop applications that operate correctly in the presence of failures and concurrency.

ACID transactions are a classical tool that makes it easier to develop applications that operate correctly in the presence of failures and concurrency.

This lecture is about using consensus to implement state machine replication. It starts by defining the problem and presenting a simple solution for synchronous systems.

We will give a survey of recent work on lattice-based cryptography, mainly focusing on the so-called Learning with Errors (LWE) problem.

Decentralized approaches spanning multiple administrative domains (MAD) are an increasingly effective way to deploy services.

Decentralized approaches spanning multiple administrative domains (MAD) are an increasingly effective way to deploy services.

Current generation Web applications run at scales of hundreds of millions of users, each with thousands of objects, requiring petabytes of data storage, even excluding large objects such as images

Decentralized systems based on distributed hash tables (DHTs) have received a lot of attention during the past decade, and systems based on this technology are now in production use.

Writing concurrent programs is notoriously difficult, and is of increasing practical importance.

This lecture is about implementing Byzantine fault tolerant state machine replication.

Naiad is a distributed system that supports efficient incremental and iterative data-parallel computation.

Decentralized systems based on distributed hash tables (DHTs) have received a lot of attention during the past decade, and systems based on this technology are now in production use.

Decentralized systems based on distributed hash tables (DHTs) have received a lot of attention during the past decade, and systems based on this technology are now in production use.

Writing concurrent programs is notoriously difficult, and is of increasing practical importance.

In this lecture IΓÇÖll introduce some of our recent work on asynchronous C (ΓÇ£ACΓÇ¥) ΓÇô a technique for making it easier for programmers to write high-performance servers and other

Concurrent programming errors arise when threads share data incorrectly.

Writing concurrent programs is notoriously difficult, and is of increasing practical importance.

We present a new class of Byzantine-tolerant State Machine Replication protocols that we call Byzantine Chain Replication.

Barrelfish is a new research operating system developed by ETH Zurich and Microsoft Research.

Barrelfish is a new research operating system developed by ETH Zurich and Microsoft Research.

Distributed Computing is the backbone of most modern internet-scale services and forms the basis for their high availability and scalability.

Writing concurrent programs is notoriously difficult, and is of increasing practical importance.

Communication primitives are an important consideration when developing distributed algorithms and systems.

Communication primitives are an important consideration when developing distributed algorithms and systems.

Communication primitives are an important consideration when developing distributed algorithms and systems.

In this talk I will introduce some traditional problems in distributed systems and describe simple algorithms to solve them.