BusinessObjects Board

How do you monitor your Business Objects XI deployment?

general discussion
#1

When developing a monitoring strategy for any enterprise level software suite, there are obviously many different things to consider. This can range from the number of users, levels of administration required, other products that the software interacts with, etc…

I have read through many threads on BOB regarding monitoring solutions ranging from full monitoring suites to PERL based scripts for monitoring services. As part of an initiative at Business Objects, I am tasked with writing a solution guide for developing a monitoring solution for the Business Objects Enterprise Platform. As a well established community of Business Objects Administrators, I am curious as to what each of you do to monitor your and your customers implementations of XI (sorry, the guide will only focus on XI r2 & XI 3.0 - no 6.5 deployments).

With this said, I would like to know:

  1. What system level components do you monitor (network, disk, cpu, etc…)?

  2. What Business Objects services do you monitor? What metrics are you monitoring for those services and what are your defined thresholds?

  3. Which 3rd party products are important to monitor as part of the overall reporting solution (databases, web application servers, SANs, etc…)?

  4. Would you prefer to see a solution from Business Objects that integrates with 3rd party products or a full solution that is embedded within the Business Objects products? Why?

  5. Anything else you want to tell me.

Please keep in mind that I am looking for input for a solutions guide to be developed by Business Objects. I can only comment on what will go into this guide during the first iteration and what we are thinking for the next iteration. In most circumstances, I cannot comment on future product direction.

cyonce :us: (BOB member since 2002-08-20)

#2

Coy, its great that you are asking the user community these questions.

I’d like to see a full embedded BO solution as I feel this is something that should be a part of your core product suite, coming free with Premium, Professional or whatever the offering is.

I’m really surprised the guide will include Xir2, I would have expected this to be XI 3.0 only.

Nick Daniels :uk: (BOB member since 2002-08-15)

#3

Although Tomcat is included in the BOE XI install package as the default web container, it is not considered as part of the XI core product. Currently, I have to use some open-source software to monitor it. It will be nice if BOBJ can provide a single-point solution to monitor Tomcat sessions, memory usage, IP addresses of incoming traffic, etc.

It will also be great if we can see all the servers’ metrics at one glance at real time, like in a table format. Currently, we have to go to CMC to click on each server to view the metrics. And the information is not real time.

That’s all I can think of off the top of my head. Thank you very much for asking!

substring :us: (BOB member since 2004-01-16)

#4

@Nick Daniels
Thanks for the feedback. For the initial release, we are looking at providing a solutions guide that will provide insight into the monitoring strategy for a successful monitoring solution for BOE. It will also provide a list of metrics that are important to measure along with their associated thresholds. A second deliverable, in addition to the guide, will be a set of SDK based probes that will allow for direct monitoring of our services.

So, we are approaching the need from a short-term / long-term mentality. In the short-term, we want to start providing a solution that will tie into other monitoring tools and even apply to those customers that do not have any monitoring tools at all. In the long-term, we are considering how we can further improve this to allow better monitoring within the product itself.

Also, as we are looking to provide something in the short-term, we feel it is important to include XI r2 as most customers are still running XI r2 rather than XI 3.0.

@substring
I agree that it is important to consider applicable 3rd party products in to the monitoring solution. There is no way to monitor the full reporting solution without including monitoring of the web servers, application servers, jvm, database and even the storage areas for the FRS. Thanks for the validation in that respect.

With regards to providing a one-stop shop for viewing all metrics. This will not be available with the first release (as you can see from my note to Nick above); however, it is something that we are looking into further for an upcoming release.

Thanks to those that have responded thus far. I look forward to getting more feedback.

Coy

cyonce :us: (BOB member since 2002-08-20)

#5

This is interesting as we are in the process of migrating to XIR2 - our infrastructure is in place and users will be migrated on to it in the next couple of weeks.

We are currently looking at what aspects of the Business Objects estate to monitor and have decided that we are interested in the entire end to end experience for a user.

Coy you mentioned monitoring discret parts of the environment but not the end user experience. I know that this is affected by many factors - but it would be useful to give a high level view of all of these together.

With this is mind we are hoping to develop a method of bench marking how long a given report takes to refresh at a remote location. ie what is the user experience like, when they are 300 miles away from the database, cms etc…

This should give us an early indication of when things are slowing down or the system is starting to struggle. If many aspects of the system suffer a minor degredation of performance this might not trigger an indiviudal alert (Database or SAN) but if a number of problems occur at once the affect may be escallated dramatically.

If we could monitor this refresh time - this would act as a general health check for the infrastucture, hopefully alerting us to problems before the users tell us. Conversely it would also help us to decide how sever a problem was - ie if we lost half of our webi report servers over night - does that affect the smaller user community at that time.

Currently in version 6 we only know we have a “performace” issue when users phone us up - we would like to be more proactive. We usually find that by the time we get end to end perfromance monitoring in place - the problem has gone.

I hope my random thoughts make sense.

Any advice that you could offer on what to monitor and how / best practises we would find really useful.

Regards

simon_cse :uk: (BOB member since 2007-06-29)

#6

things for monitoring XI

Bring back the BCA admin console. That is a big hole in the XI platform.

Better control on scheduling.

Built in server dashboard monitoring memory CPU connections.

Fix the bug that tosses warnings in event viewer on windows servers; message about transport error that customer service says to ignore.

For the cost of Bobj we as customers shouldn’t have to go looking for 3rd party tools to do things the software should do.

Mike Murray :us: (BOB member since 2005-12-23)

#7

You are correct, Simon, in the importance of monitoring the behaviour of the system from an end user perspective.

The way that I approach monitoring, and the way that the initial version of the guide approaches it, is to define your monitoring strategy according to the following process:

Define your goals
Your goals should map to the business need that the system or application is fulfilling. For example, I want to monitor login time to ensure that a login for this application on this system does not take longer than 6 seconds.

Group your goals into monitoring categories
The 4 major monitoring categories are: reliability, availability, functionality and performance. Reliability is how predictable a system is. Availability is determining whether a system is responding to requests. Functionality determines whether a system is providing the expected functionality. Performance is whether that functionality was achieved within an expected timeframe.

Define the criticality of each goal
Criticality can be defined as a minor risk, moderate risk or major risk. My example of login time can be minor, moderate or major depending upon the business need fulfilled by your application. If individiuals on a warehouse floor are making shipping decisions based on the application, I would consider login time, and session timeouts for that matter, a major risk.

Sepcify which type(s) of monitoring applies to that goal
When it comes to monitoring, the different types (or implementations) can be classified as white box, black box or system level. White box mointoring can only occur when you know how the application works internally. These are things like probes or logical component monitoring groupings based on provided functionality. Black box monitoring occurs when you have no idea how the application works internally. This is where your typical end user behaviour monitoring comes from; synthetic user transactions. System level monitoring is done by monitoring system resource usage (CPU, RAM, network bandwidth, logs, etc…)

Map your goals to system / application components
Once you have all of the above done, then you should map your goals to the system / application components. When I am logging into the Business Objects Enterprise, which components are used? The answer: system resources, web server, application server, CMS, CMS database, SiteMinder (potentially) & LDAP directory / Active Directory (potentially). Each of these components need to be monitored to create a monitoring strategy that will service your goal of monitoring login times.

Implement your monitoring strategy
Finally, you implement your monitoring strategy with home grown tools, open source tools (i.e. Nagios) or a full monitoring suite (i.e. Tivoli).

I hope this helps. The guide will go into more detail and will provide sample monitoring implementations.

cyonce :us: (BOB member since 2002-08-20)

#8

Thanks for the feedback Mike.

You bring up good points in terms of being able to not only monitor the end user interactions and the system level / process level performance, but also how schedules are behaving.

Mike: Bring back the BCA admin console. That is a big hole in the XI platform.
Coy: Have you looked at the new CMC for XI 3.0? Some aspects of the BCA console and Supervisor were brought back for the enhanced version of the CMC.

Mike: Better control on scheduling.
Coy: Again, the new CMC for XI 3.0 will provide some of this, but we are also very interested in gathering feedback from customers on what improvements they would like to see. I suggest that you provide your thoughts to us via a case logged with Technical Support stating what you would like to see in terms of improvements and the business need that it would help you fulfill. Support will do the research on whether the requested functionality is coming with a future release and if not, can log an enhancement request to request it.

Mike: Built in server dashboard monitoring memory CPU connections.
Coy: This is something that I am hearing quite a lot and is coming as a surprise to me. When starting this exercise, I expected to find that customers would want us to integrate with already existing monitoring solutions rather than building a new solution. The first release of our solution guide will provide the monitoring strategy and how to implement it regardless of whether you have an existing monitoring tool or not. It will also provide an appendix on how to implement that strategy with some of the major monitoring tools in the market. We are looking at how we can further improve our monitoring solutions for future releases of our guide (and associated tools/utilities).

Mike: Fix the bug that tosses warnings in event viewer on windows servers; message about transport error that customer service says to ignore.
Coy: I cannot really comment on this in terms of what the bug is or when / if it will be fixed. I have worked with customers that have experienced this and know that we are making improvements to get ready of erroneous error messages or improve our messaging so that you get more detailed and specific messages.

Mike: For the cost of Bobj we as customers shouldn’t have to go looking for 3rd party tools to do things the software should do.
Coy: Again, great feedback. As I mentioned above, I have been surprised at hearing this, but have definitely heard this from many people. As such, we will be taking this into consideration for future iterations of our monitoring solutions.

cyonce :us: (BOB member since 2002-08-20)

#9

Thanks for the info Coy - it sounds like a very good strategy.

For what its worth I agree totally with Mike - we had to buy a third party scheduling tool to manage our scheduling in XI. We have raised the issue we had with Business Objects and they have told us, as you have, that some of the scheduling holes have been plugged in XI3.0 - but unfortunately not all.

Monitoring the performance of the scheduling would be fantastic - I would love “something” to tell me the most efficient timings and sequence to trigger off my reports based on my hardwear, report deadlines and file watching events I have set up. Surely most if not all of the information is captured by auditor along with how “hard” the servers are performing at any one time.

Best of luck with your paper.

Regards

simon_cse :uk: (BOB member since 2007-06-29)

Monitoring Scheduler
#10

Good afternoon all.

I wanted to let each of you know that the monitoring addon for XI r2 is available for download. You can download the package with documentation and SDK-based probes from here: http://labs.businessobjects.com/enterprise_monitoring/default.asp

The monitoring guide is also available for download alone at this location: http://help.sap.com/businessobject/product_guides/boexir2/en/xir2_monitoring_guide_en.pdf

If you have any questions, comments or suggestions for improvement, please send me an email at coy.yonce@sap.com or submit them to the labs.businessobjects.com site where the package is hosted.

Thanks,

Coy

cyonce :us: (BOB member since 2002-08-20)

#11

I am unable to get to either link

:hb:

Mike Murray :us: (BOB member since 2005-12-23)

#12

Try the ones below. There were periods at the end of the original URLs that I posted.

Monitoring Addon Package:
http://labs.businessobjects.com/enterprise_monitoring/default.asp

The monitoring guide is also available for download alone at this location:

cyonce :us: (BOB member since 2002-08-20)

#13

I made the same change to the original post. Those links now work as well.

Dwayne Hoffpauir :us: (BOB member since 2002-09-19)

#14

I saw the monitoring guide last week. It is quite interesting, there are specific sections for Tivoli and MOM (think that is the correct acronym).

Nick Daniels :uk: (BOB member since 2002-08-15)

#15

The monitor is a good step forward but it uses probes to track various components. Will there possibly be monitor tools that can in real-time show what a particular service is processing? Like for example, can you watch the report/query traffic in or passing thru the WEBI processing services as things are happening?
We would like a tool or method to show/chart the jobs processed by a service, by times of the day. We don’t have a good way to show the true traffic thru our BOXI R2 system.

berzajsh :us: (BOB member since 2005-10-03)

#16

Only just picked up on this thread, have you not seen;


or

or
http://appfluent.com/businessobjects/

ABILtd :uk: (BOB member since 2006-02-08)

#17

  1. What system level components do you monitor (network, disk, cpu, etc…)?
    Particularly: Disk space and CPU.

  2. What Business Objects services do you monitor? What metrics are you monitoring for those services and what are your defined thresholds?

  • report services(webi mainly), number of requests do not hit defined limits

  • of users logged on at any one time on the CMS.

  1. Which 3rd party products are important to monitor as part of the overall reporting solution (databases, web application servers, SANs, etc…)?

  • Use Tivoli to monitor app servers and boe servers.

  • Databases are monitored seperately by the DB owners, however I set up some notifications, so that database loads status are displayed to BOXI users.(so they know when their data is available for their reports).

  1. Would you prefer to see a solution from Business Objects that integrates with 3rd party products or a full solution that is embedded within the Business Objects products? Why?

I’d like some embeded monitoring, perhaps sent via sms when services hit their defined limits in the CMC.(eg report services maxing out on requests), this level of granuality I cannot get from tivoli and would work better imbedded. this is answer in part to question 2.

tommy_b :australia: (BOB member since 2007-03-20)

#18

Hi Coy,

I know this is an old thread, but I just came across it. I downloaded your guide and noticed there’s no mention of the ‘System Connections’ metric found in the Manage Settings area of the CMC. I cannot find any documentation that explains what this setting represents or how it is controlled. Can you point me in the right direction to finding this information? Thanks!

BOSleuth (BOB member since 2005-05-05)

#19

where can I download the monitoring package ? It is not available at the mentioned link.

thanks

smeagolfer (BOB member since 2010-06-23)

#20

There is a “Get it Now” button at the bottom of this page.

ChrisW1204 :us: (BOB member since 2011-04-21)