Mishra, Anik (2011) Automated AMF Configuration Difference Generation. Masters thesis, Concordia University.
|PDF - Accepted Version|
Many domains require computer clusters to meet clients’ service level expectations. As cluster size increases, component failure becomes more likely. Rapid failure recovery is required to maintain high availability. The SA Forum has created specifications enabling management of multi-vendor solutions. These long-lived systems require tailored upgrade campaigns to keep their configuration up-to-date.
Passed works on SA Forum’s Availability Management Framework (AMF) have created an automated AMF configuration generator and upgrade campaign generator. However, to generate an upgrade campaign for an already configured cluster based on a new target configuration, a configuration difference generator is needed. Furthermore, while Distinguished Names (DNs) usually uniquely identify object in a configuration, configuration generators do not guarantee that DNs in the new configuration will match. DN modification is not possible in a system without object replacement, causing service loss. Non-DN based inter-configuration object association is needed to restore old DNs.
Our objective is to devise a technique to perform difference generation with limited knowledge of DNs and to find what data is needed to achieve this. To accomplish this, we analyse the AMF configuration model, then propose assumptions in a top down approach based on maintaining service availability during upgrade.
We propose a two phase technique that first associates objects in-between configurations and then outputs the differences. We have implemented a tool that implements this technique on the Eclipse Platform and integrates with MAGIC’s Automated Upgrade Campaign Generator. We then present a case study based on the PHASE example.
|Divisions:||Concordia University > Faculty of Engineering and Computer Science > Electrical and Computer Engineering|
|Item Type:||Thesis (Masters)|
|Degree Name:||M.A. Sc.|
|Program:||Electrical and Computer Engineering|
|Date:||06 January 2011|
|Thesis Supervisor(s):||Khendek, Ferhat and Toeroe, Maria|
|Deposited By:||ANIK MISHRA|
|Deposited On:||08 Jun 2011 14:33|
|Last Modified:||08 Jun 2011 14:33|
Repository Staff Only: item control page