Interactive Dependency-based Watchdog Methodology for Loosely Coupled Application Servers
Publication Date: 2016-Oct-09
The IP.com Prior Art Database
Disclosed is a methodology for achieving high availability between multiple loosely coupled and interdependent application servers using intelligent failure responses and indirect interactivity to avoid single full application restarts.
Page 01 of 5
The problem of high availability has many solutions including redundancy, intelligent server placement, and the use of watchdog systems. These methods function separately at different points of a distributed computing environment 's lifecycle to avoid potential failures and enforce reliability at runtime. The runtime case being handled by a watchdog system is a hardware or software layer system that provides monitoring and recovery functions for individual system processes.
When many interdependent applications must recover from an individual application 's failure, the development of a watchdog system that handles this becomes a challenge. The watchdog system must account for application level queries, response-action mapping for recovery, state-based alternate recovery methods, and decoupled synchronized subsystem monitoring. A single application server's failure will not cause the direct failure of other application servers running on the machine .
The novel solution is a methodology for achieving high availability between multiple loosely coupled and interdependent application servers using intelligent failure responses and indirect interactivity to avoid single full application restarts.
The solution includes the following methods:
1. Define a list of all application servers or an accessible directory of servers following a naming pattern, response codes from each application, and a function to execute for each response code. Create a storage system that can be accessed by applications externally from the watchdog that can store predefined flags or encoded values representing commands or states for the watchdog to apply.
2. By defining a directory of servers that follows a naming pattern, the watchdog will monitor a dynamic set of servers in case of additions during runtime.
3. The main thread of control of the watchdog is run as an operating system level service to centralize the point of failure to the operating system and to execute at system start. This will start all application servers in a user-defined order of dependency.
4. The main thread mentioned in method 3 begins with an input scan; the main thread will scan some storage; a specified directory or database for input to alter settings or perform immediate functionality that must take place before the application monitoring section of the life cycle.
5. The storage mentioned in method 4 is used as an indirect communication interface with the watchdog that allows external applications, monitored applications, or users to manually induce a state or command unto the watchdog prior to the application monitoring section of its life cycle.
6. The main thread mentioned in method 4 will begin its cycle by querying the list of application servers for their statuses. The functions defined in method 1 will be executed when a response code from the paired application is returned.
--based Watchdog Methodology for Loosely Coupled...