You’ve had your SAP BusinessObjects deployment running for awhile now. Perhaps you’ve been around since the WebIntelligence 2.x days or maybe you’ve only been running since BI 4 was released. Either way, I’m sure you’ve had a few end users problems and probably a few stability issues to resolve, but things are probably going well overall.
When EV Technologies is working with customers that fit the above scenario, the typical questions that arise are:
- How do I ensure that my system continues to run smoothly?
- How can I get ahead of end user and system issues before users start screaming?
- How do I keep my users from destroying me?
In addressing these types of questions, we tend to look at a few aspects of a SAP BusinessObjects deployment.
Routine Clean-Up of Unused Content
Any system that has been running for six months or more is bound to have some content that can be cleaned. This could be caused by historic instances, content that you thought would be useful but isn’t, or reports that your users have created in their inboxes. In any case, it’s important to keep on top of this useless content so that you can keep your BI content fresh, your deployment lean, and ensure information is easy to find for your users.
Using our Sherlock Inspector Suite, we can easily identify this content. For example, the chart below shows the number of reports by Folder Type that have never been viewed. These numbers are aggregated across a couple of our customer systems with the end result being nearly 170,000 individual pieces of BI content stored in public folders that have never been viewed. The number drops to nearly 31,000 for Inboxes and almost 21,000 for Favorites.
Going a bit further, the below tree map shows the top offenders in terms of unread content stored in Inboxes specifically. With almost 4,500 unread reports, it seems that our customers either create lots of useless content as the Administrator or send unread instances to that user. The numbers drop as we look at each of the other Inboxes displayed in the graph with the next offender having close to 900 unread reports in their Inbox.
We could take this further by showing the specific report name, full folder locations, the date the reports were created, whether they were instances, the last date a report was viewed (for those that have been viewed), the last date a report was refreshed, the date it was last edited, the owner, the size of the content, and many other details.
The end result is that you can gets lots of detail about the content that’s just sitting there wasting space. You can then drill further into those details to determine, by comparison, which content is used frequently. Thereby giving you the means to determine why some content is useful and other content is not useful.
Managing all of the scheduled jobs within an SAP BusinessObjects deployment can be a full time job. You are either implementing new schedules, managing existing schedules, tracking down errors, ensuring that things are running when they’re supposed to, or cleaning up old instances that are no longer required. Unfortunately, all of this is necessary – whether you have the time for it or not. Your users depend upon getting information when they need it and scheduling is a very useful means of achieving this goal without having ad hoc users crippling your system.
When it comes to implementing new schedules or managing existing schedules, one important aspect is ensuring that your deployment can handle the number of schedules that are running. An example of a metric that is useful for performing this analysis is shown below. In the chart, you can see the number of schedules that typically run during any given hour in the day for a deployment. As you can see, this particular deployment has just over 200 schedules running at 2PM with the next largest number being just over 70 schedules running at 2AM.
Beyond the obvious demonstrated need to spread schedules out a bit for this deployment, we can use this information to determine whether the deployment can handle this load or increase load based on the existing system configuration.
It is also important to look at the increase or decrease in the number of errors or in the time it takes for schedules to complete. The information below comes from one of the pre-built Sherlock reports. It allows you to quickly determine how many schedule failures are occurring month over month. In addition, you can also quickly determine how long schedules are taking to run each month.
There are many ways to extend this information; however, one way that we typically use scheduling information is to alert SAP BusinessObjects administrators and end users about the number of schedule failures that have occurred in the past 24 hours. We have created a pre-built report that is scheduled to be sent to administrators (for the entire deployment) and to each end user that has had a schedule failure (just for their schedules). The administrator can act on those failures as the report provides the error message for the failure. The end user becomes aware of the failure which lets them know why they haven’t received their scheduled report yet.
System Tuning Based on Real Trends
Even for a deployment that seems to be humming along pristinely, it is important to track trends in content growth, user community expansion, schedule utilization, and other areas. This allows you to ensure that you keeping your system tuned for real usage and not some estimate of growth. In addition, it allows you to have a real picture of how large your system is for when you are talking about expanding your usage of SAP BusinessObjects across your organization.
By tracking information about the growth of your BI community, you can determine exactly how many users your deployment is servicing. You should further break this information down into active versus inactive users, types of users (e.g., viewers, editors, creators), frequency of logins from active users to more concretely determine the impact these users may be having on your system.
The same would be true for the amount of content being created month over month or the number of schedules created month over month. By tracking these metrics, you can determine how much load the active users are actually putting on the deployment. You can also further break this down into types of BI content, BI versus agnostic content, number of new reports versus edited reports, etc…
By combining the information above with details about the types of BI services deployed and the current configuration for those services, you can get a very good idea of how much your deployment can handle. This will allow you to determine if you can handle the load or if you need to modify your configuration.
Proactive Error Awareness
Another area with which you should be concerned is the continual monitoring of errors that occur within the environment. Ideally, you will get ahead of these errors before your users are even aware that they are a problem. This is true whether it is a problem with one report or with the entire system.
With our customers, we implement a set of reports that continually track errors that are occurring with users who are logging into the SAP BusinessObjects platform to access content and with schedules that are running. We even deliver a pre-built report with Sherlock that provides this information as well.
Reducing Instability Caused by Content
Finally, it is important to keep track of any content that may potentially cause a problem with the deployment. For example, you will want to keep any eye on any schedules that take longer than what is normal for your deployment or reports sent to the scheduler that are larger than a certain size. In addition, you may want to keep an eye on the size of reports created in your deployment to ensure your users aren’t creating massive content that the BI system then needs to process.
With all of the above combined, you will have a good start to implementing a strategy for ensuring that your SAP BusinessObjects deployment remains successful. By keeping user errors in check, ensuring that your deployment is tuned to handle real capacity, and monitoring content usefulness; you can keep your BI community happy.