Erlang’10 Proceedings of the 2010 ACM SIGPLAN Erlang by General Chair: Scott Lystig Fritchie, Program Chair:

By General Chair: Scott Lystig Fritchie, Program Chair: Konstantinos Sagonas

Show description

Read Online or Download Erlang’10 Proceedings of the 2010 ACM SIGPLAN Erlang Workshop PDF

Best networking books

802.1X Port-Based Network Access Authentification

Written to attract a wide viewers, 802. 1X Port-Based Authentication seeks to outline this complicated notion in obtainable phrases and to discover its quite a few purposes to contemporary laptop networks utilizing this actual community protocol. this article assumes that the reader can have very little previous wisdom and just a normal knowing of desktop networking, providing an introductory evaluation in addition to a subject matter evaluate in every one bankruptcy.

Additional info for Erlang’10 Proceedings of the 2010 ACM SIGPLAN Erlang Workshop

Example text

However, given the problems that we know have happened in production networks, it is likely that a distributed manager application would likely have been fooled by the same conditions and made similarly bad decisions. Detecting Brick Failure and Network Partition The original chain replication paper [24] makes two assertions that are extremely problematic in the real world. ” If either assumption is violated, the system can quickly make bad decisions that can cause data loss. The biggest problem with detecting halted nodes is the problem of network partition.

Fredlund. Programming distributed Erlang [7] H. Svensson and L-A. applications: pitfalls and recipes. In Proc. of the SIGPLAN workshop on Erlang, pages 37–42, New York, USA, 2007. ACM. The built-in registry is another difference between the current and the new semantics. However, its workings should be easily made equivalent to the standard process registry implementation in Erlang. Also, because of the introduction of a general treatment of side-effecting actions, the inclusion of a registry affects the size of the semantics marginally.

Partitions are usually caused by failure of network equipment, such as an Ethernet switch or failure of network links such as a telecom data circuit. However, any failure of hardware and/or software that creates arbitrary message loss can be considered a network partition. With message passing alone, it’s impossible to tell the difference between a network partition, a failed node, or merely a very slow node. The built-in Erlang/OTP message passing and network distribution mechanisms cannot adequately handle network partition events by themselves.

Download PDF sample

Rated 4.65 of 5 – based on 31 votes