Distributed naming in a factored operating system

A factored operating system (fos) is a new operating system design for manycore and cloud computers. In fos, OS services are separated from application code and run on distinct cores. Furthermore, each service is split into a fleet, or parallel set of cooperating processes that communicate using messages. Applications discover OS services through a distributed, dynamic name service. Each core runs a thin microkernel, and applications link in a user-space library called libfos that translates service requests into messages. The name service facilitates message delivery by looking up service locations and load balancing within service fleets. libfos caches service locations in a private cache to accelerate message delivery, and invalid entries are detected and invalidated by the microkernel. As messaging is the primary communication medium in fos, the name service plays a foundational role in the system. It enables key concepts of fos's design, such as fleets, communication locality, elasticity, and spatial scheduling. It is also one of the first complex services implemented in fos, and its implementation provides insight into issues one encounters while developing a distributed fos service. This thesis describes the design and implementation of the naming system in fos, including the naming and messaging system within each application and the distributed name service itself. Scaling numbers for the name service are presented for various workloads, as well as end-to-end performance numbers for two benchmarks. These numbers indicate good scaling of the name service with expected usage patterns, and superior messaging performance of the new naming system when compared with its prior implementation. The thesis concludes with research directions for future work.

Department dc:contributor.department

Massachusetts Institute of Technology. Dept. of Electrical Engineering and Computer Science.

Grantor dc:publisher

Massachusetts Institute of Technology

Year dc:date.issued

2010

Author dc:creator

• Beckmann, Nathan (Nathan Zachary)

Advisor dc:contributor.advisor

• Anant Agarwal.

• Electrical Engineering and Computer Science.

Statement dc:rights

• M.I.T. theses are protected by copyright. They may be viewed from this source for any purpose, but reproduction or distribution in any format is prohibited without written permission. See provided URL for inquiries about permission.

Licence dc:rights.uri

• http://dspace.mit.edu/handle/1721.1/7582

Language dc:language.iso

eng

Handle dc:identifier.uri

http://hdl.handle.net/1721.1/62460

OAI identifier oai:identifier

oai:dspace.mit.edu:1721.1/62460