I posted a tweet this morning about a decision I had to make this week that was not easy for me, and several requested some additional information to clarify. Let it be known, that I, Greg Myers, SAP Mentor, ASUG Volunteer, and BusinessObjects Certified Professional, recommended we use the Microsoft Reporting and Analysis Services for an upcoming project. (insert collective gasp here).
Quick background on the project. I’m working on building an internal data repository for all of our server and application performance metrics. It’s going to be huge in a few months. Our goal is to use this data to be able to forecast capacity of our massive “platform”, to model what types of business activities have what effects on our infrastructure, and to forecast what adding new clients (and what types of clients) will do to performance and capacity. It’s a tall order.
We have performance monitoring tools that gather many of the statistics for us, but they are light on the reporting side, so we’ve been looking at other ways to get to where we need to go. My first idea was to use an Oracle database for our back end, and use BusinessObjects on the front side for reporting. When we began doing our breakdowns of what that would actually entail, it quickly became a sub prime option (I will do some comparisons in a moment). The next option we considered was a Microsoft SQL Server back end, with BOBJ on the front side for reporting. And the third option (probably final) was to use Microsoft thru and thru (SQL Server, Analysis and Reporting Services). And yes, I chose the final option. Here’s why:
1. An Oracle back-end requires lots of technical know-how to build, manage, tune, and maintain. We have several Oracle DBA’s, but none have the bandwidth to take on an extra project like this. I cannot support Oracle by myself.
2. We need OLAP cubes to do some of the correlation and modeling we are proposing. Oracle can do them, but again require lots of brain power (that I don’t have).
3. Since we need OLAP cubes, WebIntelligence isn’t really the proper tool to go browsing through those. SAP has the tools, but we aren’t currently licensed for them.
4. Much of the metrics will be going into executive dashboards. We want to be able to integrate with our corporate SharePoint site since that is where our managers are used to going for information. We can’t currently integrate BOBJ with SharePoint because 64-bit SharePoint isn’t supported with the Integration Kits.
5. None of the solutions I’ve discussed so far have any type of ETL tool that we are licensed for or have at hand to use. We have to move the data somehow.
6. We have a large SQL Server farm which is licensed for Analysis and Reporting Services, and staff that has the ability to support a new warehouse of this size.
7. The Microsoft BI stack integrates very cleanly with our current version of SharePoint.
8. The Microsoft BI stack (and SQL Server) includes an ETL Tool (SSIS)
9. The Microsoft solution will cost us roughly one fourth of what the Oracle/BOBJ solution would have cost.
In the end, it was purely a business decision. We needed something that fit into our current landscape that would be easy to support and maintain, and not break the bank to implement. I certainly have my personal preferences, and I know that the Oracle/BOBJ solution would have been way more sophisticated, but I’m not sure sophistication is needed in this case. In the end, I am happy with my decision and recommendation, because I truly feel it’s the best solution for this problem.