Skip to content

Latest commit

 

History

History
63 lines (55 loc) · 3.48 KB

README.md

File metadata and controls

63 lines (55 loc) · 3.48 KB
Raw

High-Availability-Processes

Implementation of client, proxy, and server with high availability:
Fault-Tolerant Tuple Space Service
The specifications are in specs.doc
this spec where written by my Dr. Phil Pfeiffer.

Details

CSCI 5150: Phil Pfeiffer
Software-based final exam – fault-tolerant tuple space service
Terms: As individuals or in pairs
Due date: Monday, 10 December, close of business, with interviews to follow

Background:

One reason for implementing a distributed system is to assure a service's fault-tolerance: i.e., its ability to operate after component failures. One classic architecture for ensuring fault tolerance, the active-passive architecture, uses a second, hot-swap server to back up a primary server (Figure 1). This configuration operates as follows:

  • The primary (i.e., active) server, in normal operation,
    • Receives and responds to all requests from client computers
    • Relays all requests for updating the primary's internal state to the backup, which maintains a duplicate of the primary's internal state.
  • The primary and backup servers, in normal operation, monitor each other's status, using "heartbeat" messages.
  • If one of the servers detects that the other has failed:
    • If the primary detects a failure of the backup
      • It notifies an operator that a failure has occurred and keeps operating.
      • The operator should then
        • restore the backup
        • initiate a procedure for resynchronizing its state with the primary server
    • If the backup server detects a failure of the primary
      • It triggers the computer-controlled switch, rerouting of client requests from the failed primary to itself.
      • It notifies an operator that a failure has occurred and keeps operating.
      • The operator should then
        • restore the primary, which becomes the backup
        • initiate a procedure for resynchronizing its state with the primary

This assignment asks you to implement a comparable configuration, using a proxy server in place of the "deadman" switch (Figure 2).

Requirements:

The requirements for this assignment are divisible into requirements for

  • The service
  • Primary and backup servers;
  • Proxy server
  • Client application

Functional Requirements for Service

  • The service shall provide a database service in the form of a tuple space: i.e.,
    • a collection of (key, value) pairs, where
      • keys and values are n-tuples of alphanumeric strings
      • the space has at most one value per key
    • together with four operators:
      • GET - given a (keyexp, valexp) pair,
        • if the (keyexp, valexp) pattern is malformed, return an empty list. Note: assume Python regular expression syntax for keyexp and valexp.
        • else, return all (key, value) pairs such that
          • key matches keyexp
          • value matches valexp
      • PUT -given a purported list of (key, value) pairs,

        • if the list is malformed, return an indication to this effect

        • else

          • for every pair such that
            • key and value are well-formed and
            • key is not yet present in the tuple space

          add (key, value) to the tuple space The specification (specs.doc) and was written by Dr. Phil Pfeiffer.