Apr 15, 2023

Application Benchmarking of a siebel system for a leading US based CDN Cloud Service Provider

Posted by

Client Background

Client is a U.S. based leading content delivery network service provider (CDN)and cloud services provider.

Business Context

Client was using Siebel 7.8 for their internal account, contract and ticket management functions. Owing to the various performance issues that they faced in the past, Oracle recommended them to upgrade to Siebel 8.1. However, before they rolled out to production, they realised that the performance issues were still prevalent in v8.1 as well.

How did we help the Client?

Given the proprietary nature of Siebel, the kind of forms used and the protocols underneath, Ticking Minds recommended NeoLoad as the solution for the performance testing – since NeoLoad supports Siebel out of the box. Following was the approach taken:

  • Critical business transactions were identified based on the business priorities, past production logs and frequency of transactions carried out. A workload model was drawn up to identify the transaction mix amongst the 15 business critical transactions identified, some of which included web services as well. Nagios was chosen as the monitoring solution across the physical layers since the client had deployed this solution across the organisation for a while now. 4 geo locations were identified – Cambridge USA, Krakow Poland, Santa Clara USA and Bangalore India – from where the load was expected to be injected on to the server. SLAs were identified in terms of response times, server CPU and RAM utilisation thresholds etc.
  • Test scenarios were recorded using NeoLoad with Siebel plugin in a seamless manner and were correlated & parameterised as appropriate. On-premise load generators were set-up using NeoLoad and they were set-up with a central controller to manage the load from different load generators. Load, Stress and Endurance tests were conducted – several bottlenecks were identified predominantly with the indexing carried out in the Oracle 11 database tables.

Benefits Delivered

Post tuning the database, all transactions were executed within the 2-3 seconds SLA threshold while meeting all the server thresholds as well. Key performance issues were unearthed largely in the database layer, where indexing recommendations brought-in significant 30% improvement in response times. No tool other than Neotys has great monitoring capability for the Siebel platform, which had provided about 50% effort saving in monitoring and resolution.