Hitachi, Ltd.

Hitachi Software Engineering Company, Ltd.

Shigehiro Hatakeyama

Takafumi Nogami

Takafumi Jinsenji

Naoyuki Matsunaga

Toshio Hata

Takahiro Morimoto

Hiroyuki Shinozaki

Tomohiro Sakaue

Kazutoshi Amano

Fumio Gomi

Takenori Iwato

Masahiko Hashimoto

A computer network comprising a monitoring apparatus and a plurality of servers, wherein the servers respectively have a plurality of software programs, and a data file for storing error analysis information to be utilized for analysis of the cause of an error upon occurrence of error, for each software program. Further, the servers respectively have a function to transmit an error notifying message including a software identifier of a software program executed upon occurrence of error to the monitoring apparatus. The monitoring terminal apparatus has a function to instruct at least one server, specified based on the software identifier included in the error notifying message, to save error analysis information, in response to the error notifying message received from any one of the servers.

Distributed data processing system and error analysis information saving method appropriate therefor

A method and a system for analyzing the operation of a parallel data system are disclosed. The history information of the access executed by a specific processor that has received a request for access to a plurality of data bases connected to a plurality of processors, the ID of the specific processor and the accessed data amount are collected and stored in a memory. The history information is compared with the time assumed from a predetermined time specified for the access to a unit amount of data in data bases, deciding whether the former is reasonable or longer than the latter. The transfer history information for the processor that has transferred an access request to the specific processor executing the access is compared with a predetermined transfer time specified for a unit amount of data, deciding whether the transfer time between the two processors is longer than an assumed transfer time. It is thus possible to determine whether an abnormal condition of the parallel data base system is caused by an internal fault of the processors or due to a defect of the access request transfer path between processors.

Access analysis method and system in parallel data base system

Growing community of inventors

Yokohama, Japan

Shigehiro Hatakeyama