Browse Prior Art Database

Method for Column-based Historical Information System Data Access

IP.com Disclosure Number: IPCOM000200042D
Original Publication Date: 2010-Oct-12
Included in the Prior Art Database: 2010-Oct-12
Document File: 3 page(s) / 213K

Publishing Venue

Siemens

Related People

Juergen Carstens: CONTACT

Abstract

Almost all relational databases nowadays are primarily designed to handle online transactional processing (OLTP) workloads. Transactions are utilized using SQL (Structured Query Language), or SQL based procedural languages (such as PL/SQL and T-SQL). For a very read-intensive and information selective application, for example historical data-querying in historical information system (HIS), this kind of model reaches its limits for certain use-cases and influences processing performance bottlenecks in historical information system database engine. The limited RDBMS (Relational Database Management System) capability to process only certain amount of data within the given time frame is additional constraint, which furthermore influences performance. At present, HIS run-time engine is I/O (Input/Output) intensive row-based RDBMS. This is because the HIS system is designed to serve data for different types of clients. Read-processing dominates and it is typically part of the query processing triggered by the client. As the database is becoming more and more mature query read-processing time becomes dependent on the number of clients and their overall activity and data-demand, HIS job activities, and overall system performance. HIS short-term archive is moved to medium-term archive after some, in advance configured, time-interval. The medium-term archives are extracted to long-term archives after a predefined period of time and are totally dependent on external database job processing ability.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 46% of the total text.

Page 01 of 3

(This page contains 01 pictures or other non-text object)

(This page contains 00 pictures or other non-text object)

Method for Column-based Historical Information System Data Access

Idee: Dr. Dejan Dimitrijevic, DE-Nuremberg

Almost all relational databases nowadays are primarily designed to handle online transactional processing (OLTP) workloads. Transactions are utilized using SQL (Structured Query Language), or SQL based procedural languages (such as PL/SQL and T-SQL). For a very read-intensive and information selective application, for example historical data-querying in historical information system (HIS), this kind of model reaches its limits for certain use-cases and influences processing performance bottlenecks in historical information system database engine. The limited RDBMS (Relational Database Management System) capability to process only certain amount of data within the given time frame is additional constraint, which furthermore influences performance.

At present, HIS run-time engine is I/O (Input/Output) intensive row-based RDBMS. This is because the HIS system is designed to serve data for different types of clients. Read-processing dominates and it is typically part of the query processing triggered by the client. As the database is becoming more and more mature query read-processing time becomes dependent on the number of clients and their overall activity and data-demand, HIS job activities, and overall system performance. HIS short-term archive is moved to medium-term archive after some, in advance configured, time-interval. The medium-term archives are extracted to long-term archives after a predefined period of time and are totally dependent on external database job processing ability.

To get around the RDBMS bottleneck performance phenomenon and typical row-based query inefficiencies, row-based RDBMS utilize indexing, horizontal partitioning, materialized views, summary tables and parallel processing methods. For example, indexing method can help queries become faster, but indexes on the other hand require more storage and must be maintained, what actually degrades performance, when indexes become more fragmented; HIS Index Organized Tables (IOTs) method, Partitioning and Summary Tables (in form of persistent calculations) methods are established as performance improvement methods for the typical inefficiency of the row-based RDBMS query mechanism as well as the above mentioned separation of archives to save and reduce persistent storage amounts needed for the data archival.

In order to obtain the HIS data faster then it is the case with classical RDBMS, column-oriented HIS Plugin is proposed (COHIS). COHIS plugin-architecture is presented on Figure 1. The plugin contains following components: The session management component, the column-based query-processor, and column-data-persistence-management component. The session management component (see Figure
1) accepts all incoming client SQL requests on a speci...