Browse Prior Art Database

Selective diagnostic server routing of workflow requests based on failure probability

IP.com Disclosure Number: IPCOM000249330D
Publication Date: 2017-Feb-16
Document File: 2 page(s) / 74K

Publishing Venue

The IP.com Prior Art Database

Abstract

Server configurations that provide detailed failure diagnostics incur an increased processing overhead which is not acceptable in a highly responsive application. The problem is particularly acute for low-frequency or highly intermittent failures, where an expensive diagnostic configuration may need to be in place for an extended period of time to catch the failure again. In cloud environments there may be an immediate financial cost in running a diagnostic configuration that consumes more CPU time or memory.

This idea provides a low-cost method of obtaining diagnostics for a low-frequency or highly intermittent failure. In a data processing system, a router distributes work-flow requests to a set of server machines. As the server machines need to be highly responsive, with a low processing cost per request, they are configured with a minimal level of diagnostics. If the processing of a request fails, a basic error message is returned to the router. If there have been earlier failures for a request type, the request is routed to a server that has been configured with additional diagnostics, for example to log comprehensive trace data to a file as each work-flow request is processed. Diagnostics for a subsequent failure can then be obtained without increased processing cost for other traffic.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 51% of the total text.

Selective diagnostic server routing of workflow requests based on failure probability

Server configurations that provide detailed failure diagnostics incur an increased processing overhead which is not acceptable in a highly responsive application. The problem is particularly acute for low-frequency or highly intermittent failures, where an expensive diagnostic configuration may need to be in place for an extended period of time to catch the failure again. In cloud environments there may be an immediate financial cost in running a diagnostic configuration that consumes more CPU time or memory.

This idea provides a low-cost method of obtaining diagnostics for a low-frequency or highly intermittent failure.

In a data processing system, a router distributes work-flow requests to a set of server machines. As the server machines need to be highly responsive, with a low processing cost per request, they are configured with a minimal level of diagnostics. If the processing of a request fails, a basic error message is returned to the router. For example, an HTTP request may be sent to one of the server machines, which returns "Error 500 Internal server error". This message is insufficient to diagnose the problem, and little or no information about the failure is preserved in the server machine. The router logs the failure in a table, along with key information about the type of request that was being processed. For example, the HTTP request may include in its data a text field or ID indicating a 'bank account update' which can be used to identify the type of request. Information about the timing and / or frequency of the failure may be logged in the router's table. As subsequent work-flow requests arrive, the router looks up the request type for each request in a table. If the...