Electing a master server using election periodic timer in fault-tolerant distributed dynamic network systems

A method for operating an election mechanism, comprising:

• detecting, by at least one back-up server, when a master server is not functional, said master server possessing the mastership of a server group, said server group comprising said master server and said at least one back-up server, and electing, when said master server is detected not functional a new master server based on at least one election periodic timer, each of said at least one election periodic timer being associated with a different server in said server group and specifying a length of time relative to the time at which the master server is detected not functional causing said electing to occur at different times for at least some of the servers, said new master server taking over said mastership.

A fault-tolerant server group in distributed dynamic network systems, comprising:

• a client;
• and a fault-tolerant server group for providing a service to said client, said fault-tolerant server group comprising a master server and at least one back-up server, said master server communicating with said client, wherein each server in said fault-tolerant server group, including said master server and said back-up server, is associated with an election mechanism enabling said fault-tolerant server group to elect, when said master server fails, a new master server, in a process in which at least some of the election mechanisms are triggered by timers set to different times relative to the time at which said master server is detected to have failed.

A system for an election mechanism, comprising:

• a detection mechanism for detecting, by at least one back-up server, when a master server is not functional, said waster server possessing the mastership of a fault-tolerant server group, said server group comprising said master server and said at least one back-up server;
• and an election mechanism for electing, when said master server is detected not functional, a new master server based on at least one election periodic timer, each of said at least one election periodic timer being associated with a different server in said server group and specifying a length of time relative to the time at which the master server is detected not functional, causing at least some of the servers to elect at different times, said new master server taking over said mastership.

A computer readable medium having program code stored thereon, such that when the code is read and executed by a computer, the computer is caused to:

• receive, by a master server in a fault-tolerant server group, a request sent by a client, said fault-tolerant server group comprising said master server and at least one back-up server, said master server communicating with both said client and said at least one back-up server, each saver in said server group, including said master server and said at least one back-up server, having an election mechanism, enabling said fault-tolerant server group to elect, when said master server fails, a new master server, in a process in which at least some of the election mechanisms are triggered mechanisms are triggered by timers set to different times relative to the time at which said master server is detected to have failed;
• process, by said fault-tolerant server group, said request to produce a result, said request being processed concurrently by said master server and said at least one back-up server;
• and send, by said master server, said result to said client.

A computer readable medium having program code stored thereon, such that when the code is read and executed by a computer, the computer is caused to:

• detect, by at least one back-up server, when a master server is not functional, said master server possessing the mastership of a server group, said server group comprising said master server and said at least one back-up server;
• and elect, when said master server is detected not functional, a new master server based on at least one election periodic timer, each of said at least one election periodic timer being associated with a different server in said server group and specifying a length of time relative to the time at which the master server is detected not functional, causing at least some of the servers to elect at different times, said new master server taking over said mastership.