So had a bad today and need to vent…..
When doing a project there are a few things we do as a matter of course and are what we call Optimal Practices (as part of our OptimalBI company), these include:
- Reconciling each level of the data warehouse to ensure data consistency
- Use of symantic layer for all end user reporting
- Regular user workshops throughout development to ensure good user engagement and feedback
- Use of Wiki to enable interactive documentation
- Monitoring of the Production and Development servers to monitor and report performance and assist with capacity planning
So with the last one in mind we purchased SAS ITRM to enable us to monitor the SAS Servers on one of our projects.
We run a multi (n) tier architecture, so dedicated Web tier, Metadata Tier, Application Tier (Workspace/Stored Processes Servers) and Data Tier.
So ITRM being a SAS Solution, based on SAS 9.1 and utilises DI Studio and Web Report Studio, multi tier install here we go.
We apparently no.
It seems ITRM uses a custom metadata repository and as such required Web Report Studio etc to be installed/deployed separately.
In addition it comes with some prebuilt DI Studio jobs to read the adapters/log files (i.e SAR,BMC Patrol) but you have to build your own Performance Data Mart and Reports (in version 3.x that is as the old version 2.7 had lots of these out of the box).
So best practise is apparantley to install it on its own server and then build your own.
Not impressed.
Here end the rant
ps. as per my previous post I explicitly deny permission for Microstrategy to reproduce any content from this blog
pps. as if they would be reading my blog!
The upcoming SAS ITRM product release (ITRM 3.2), built using the latest SAS 9.2 components/technologies, addresses your requirements regarding “dependent repositories” (e.g. ITMS repository) and will also deliver out-of-the-box reports for supported SAS ITRM adapters. Having said that we would like you to continue your support for the product in future. If you shoot me an email, we can discuss your concerns in more detail.
Yes, I think you were being a bit harsh. Its not so hard as you make it out – the templates provide the same support for many data sources including BMC – and v2.7 did not come with prebuilt datamarts (PDBs) and reports for BMC either – but the adapater for reading the BMC data was there – and it is there in ITRM v3 too. We have had no issues with having a custom repository, and generally find it runs well with all pieces on a single tier. You would only have a seperate instance of WRS if you going to integrate this all into an existing DI setup – which I dont think is a good architectural choice anyway. Why place system performance data processing as an overhead on production systems? If your data feed breaks, ITRM would need to wait for it to arrive, in which case you have the risk of running performance processing during your production processing window. Not good. I think its a matter of expectations, and while v3.2 seems to be adressing much of that in a significant way, I still find that using ITRM sure beats writing your own. The idea of writing your own in ITRM is not like completely writing your own. ITRM comes with so much of the work already done. Even for a brand new custom data source, at most you do a one off exercise to define the metadata and then use the DI functionality to build the rest. I would always choose ITRM over the alternatives because it gives me a significant head start compared to ‘building your own’.