Recently at the Diamond Developer conference and the Peer Source Road Show I gave a presentation entitled Architecting for Performance. Part of my presentation covered sizing Business Objects Enterprise systems. Until recently the two main guides for customers looking for sizing help were the Administrators guide and a sizing guide that was only available to customers under NDA.Business Objects just released a public sizing guide for Business Objects Enterprise XI R2. This new white paper written by one of Business Objects best performance and scalability experts is a great resource for any administrator who is concerned about performance, security, and scalability.
In my five years of experience with Business Objects Enterprise systems I have always found sizing and performance to be one of the most challenging aspects of creating a ‘bullet proof’ system. This is also a common issue for web site administrators, who have the same issues of having to predict user load and frequency. Adding database queries and their affect on the system just makes this even harder.
My tip of the day for any administrator out there is to take advantage of report scheduling whenever possible. Report scheduling allows you to run the database queries ahead of time and at off peak hours. As far as scalability and performance is concerned Business Objects Enterprise has got to be one of the best business intelligence systems out there. This was confirmed by IBM in independent benchmarking tests completed on Crystal Enterprise 10 which boast the same architecture and framework as Business Objects Enterprise XI R2. Even with all of this, administrators can see huge advantages if they can remove the database query from the time it takes to serve reports.
Real-time data requirements are the biggest reason why most users resist scheduling reports, and this is a very valid reason. I would just add that in many instances where I have worked through these issues with the customer they have discovered that in reality they don’t really need real-time data, or many times after further investigation we discover that the data they intended on reporting off, wasn’t real-time. Definitely some thing to think about as you build out your own business intelligence system.