Services - System Architecture and Scalability Analysis (Application, Server, Database & Storage)
Assume that an application component functions within a "reasonable" response time goal when the table(s) are small. As the
table(s) and database size increases, the performance of the application component starts to degrade in an "exponential"
fashion. You expect "linear" degradation, but the results are surprisingly otherwise. Ring a bell?
The core problem here is that the application and/or its "environment" has inherent scalability issues. What works well with
small datasets, completely falls apart with large datasets. We pride in ourselves to be able to determine scalability issues with
such components in your system.
In Computer Science, the scalability of a code module is usually measured by its "algorithmic complexity". In simple terms, the
algorithmic complexity is a mathematical measurement of how well the code will perform. The measurement of algorithmic
complexity is directly related to the abstract data types and data structures that are utilized in the application. For example,
linked lists portray an algorithmic complexity of O(n). This means that if the linked list is comprised of "n nodes", the
algorithm manipulating may require up to "n" node traversals to get to the data that is sought after. In comparison, a binary
tree's algorithmic complexity is O(logn). Given the same number of nodes, a binary tree provides much more scalable
performance when compared to a linked list.
Scalability problems can (and will) stem from bad application design, inappropriate data structures, inefficient SQL, system
resource limits, database and schema architecture constraints and improper storage configuration. Early detection, diagnosis and
resolution of scalability problems is key for stable system performance health in the long run.
Assume that an application component functions within a "reasonable" response time goal when the table(s) are small. As the
table(s) and database size increases, the performance of the application component starts to degrade in an "exponential"
fashion. You expect "linear" degradation, but the results are surprisingly otherwise. Ring a bell?
The core problem here is that the application and/or its "environment" has inherent scalability issues. What works well with
small datasets, completely falls apart with large datasets. We pride in ourselves to be able to determine scalability issues with
such components in your system.
In Computer Science, the scalability of a code module is usually measured by its "algorithmic complexity". In simple terms, the
algorithmic complexity is a mathematical measurement of how well the code will perform. The measurement of algorithmic
complexity is directly related to the abstract data types and data structures that are utilized in the application. For example,
linked lists portray an algorithmic complexity of O(n). This means that if the linked list is comprised of "n nodes", the
algorithm manipulating may require up to "n" node traversals to get to the data that is sought after. In comparison, a binary
tree's algorithmic complexity is O(logn). Given the same number of nodes, a binary tree provides much more scalable
performance when compared to a linked list.
Scalability problems can (and will) stem from bad application design, inappropriate data structures, inefficient SQL, system
resource limits, database and schema architecture constraints and improper storage configuration. Early detection, diagnosis and
resolution of scalability problems is key for stable system performance health in the long run.
 |
| |
 |
| |
 |
| |
 |
| |
 |
| |
 |
| |
 |
| |
 |
| |
 |
DBPerfMan - In The Business of
Optimizing Oracle Database
Performance!
Optimizing Oracle Database
Performance!
| Copyright © 2004-2007 DBPerfMan, LLC. All rights reserved. |
