Business Service Management (BAC/BSM/APM/NNM)
More than everything monitoring, BSM provides the means to determine how IT impacts the bottom line. Its purpose and main benefit is to ensure that IT Operations are able to reactively and proactively determine where they should be spending their time to best impact the business. This covers event management to solve immediate issues, resource allocation and through reporting performance based on the data of applications, infrastructure, networks and from third-party platforms. BSM includes powerful analytics that gives IT the means to prepare, predict and pinpoint by learning behavior and analyzing IT data forward and backwards in time using Big Data Analytics applied to IT Operations.

HP Diagnostics - drinking our own merlot

HP Diagnostics enables you to seamlessly monitor,triage, and diagnose critical problems in composite applications—in both pre-production and production environments. This software collects performance and availability data from Java, .NET, and SAP platforms,isolating performance and availability problems and reducing mean time to recovery (MTTR).


What you may not know is that every Diagnostics server is instrumented with our own agent out of the box and provides performance metrics that our R&D team typically use to troubleshoot problems during development and production. This information is not exposed to customers directly but can be accessed via http://<Diagnostics Commander>:2006/query and clicking on “Open Diagnostics” for “Mercury System”.


Michael Haeuptle is one of HP Diagnostics senior architects .  Not only does Michael develop HP Diagnostics but he also uses the product to solve application issues.  Here’s Michael’s story:


During routine performance testing we recently found a slow server request Purger.run() which took on average 17.5 seconds in the last hour.



There was even an instance where it took almost 20s. The glyphs (triangles and circles) represent call profiles for a particular instance in time. A triangle that points upwards represents a maximum call profile, a triangle pointing downwards a minimum call profile and the circle represents an average call profile. By default, Diagnostics collects and keeps two maximum, one minimum and one average call profile for each server request per time bucket (e.g. 5m, 20m, 1h, etc).


This is very powerful! Unlike other solutions, where call profile collection needs to be configured manually, Diagnostics always has call profiles for selected server request instances available which are invaluable for troubleshooting. This is especially important in production where performance problems tend to be sporadic and a call profile is needed right then when a server request is taking a long time.


In our particular case, we drilled down to a maximum call profile and found the following call profile:



The call profile revealed that the method Purger.guessOnDiskSize is called over and over and is the main contributor to the overall time for Purger.run().


Notice the fuzzy bars in the call profile. These measurements come from the new 8.0 stack trace sampling feature that combines the instrumentation approach (where method entry and exit are injected with measurement code) with the sampling approach that causes almost zero overhead but at the expense of not always recording all invocations.  Armed with this information we circled in on the offending code and indeed found that Purger.guessOnDiskSize() was called in a tight loop for every file in the Diagnostics database.  After some refactoring and adding some caching we were able to increase the performance of the Purge.run() method.


Using our own agent for performance analysis and troubleshooting paid off! Having call profiles available all the time without having to manually configure the capture like other solutions require is essential.


This blog was written by Michael Haeuptle, senior architect for HP Diagnostics

Search
Showing results for 
Search instead for 
Do you mean 
About the Author(s)
  • Doug is a subject matter expert for network and system performance management. With an engineering career spanning 25 years at HP, Doug has worked in R&D, support, and technical marketing positions, and is an ambassador for quality and the customer interest.
  • Dan is a subject matter expert for BSM now working in a Technical Product Marketing role. Dan began his career in R&D as a devloper, and team manger. He most recently came from the team that created and delivered engaging technical training to HP pre-sales and Partners on BSM products/solutions. Dan is the co-inventor of 6 patents.
  • This account is for guest bloggers. The blog post will identify the blogger.
  • Manoj Mohanan is a Software Engineer working in the HP OMi Management Packs team. Apart being a developer he also dons the role of an enabler, working with HP Software pre-sales and support teams providing technical assistance with OMi Management Packs. He has experience of more than 8 years in this product line.
  • HP retiree Author of 42 Rules for B2B Social Media Marketing
  • Nimish Shelat is currently focused on Datacenter Automation and IT Process Automation solutions. Shelat strives to help customers, traditional IT and Cloud based IT, transform to Service Centric model. The scope of these solutions spans across server, database and middleware infrastructure. The solutions are optimized for tasks like provisioning, patching, compliance, remediation and processes like Self-healing Incidence Remediation and Rapid Service Fulfilment, Change Management and Disaster Recovery. Shelat has 21 years of experience in IT, 18 of these have been at HP spanning across networking, printing , storage and enterprise software businesses. Prior to his current role as a World-Wide Product Marketing Manager, Shelat has held positions as Software Sales Specialist, Product Manager, Business Strategist, Project Manager and Programmer Analyst. Shelat has a B.S in Computer Science. He has earned his MBA from University of California, Davis with a focus on Marketing and Finance.
  • Architect and User Experience expert with more than 10 years of experience in designing complex applications for all platforms. Currently in Operations Analytics - Big data and Analytics for IT organisations. Follow me on twitter @nuritps
  • 36-year HP employee that writes technical information for HP Software Customers.
  • Pranesh Ramachandran is a Software Engineer working in HP Software’s System Management & Virtualization Monitoring products’ team. He has experience of more than 7 years in this product line.
  • Ramkumar Devanathan (twitter: @rdevanathan) works in the IOM-Customer Assist Team (CAT) providing technical assistance to HP Software pre-sales and support teams with Operations Management products including vPV, SHO, VISPI. He has experience of more than 12 years in this product line, working in various roles ranging from developer to product architect.
  • Ron is a subject matter expert for BSM\APM, Currently in the Demo Solutions Group. Ron have over thirteen years of technology experience, and a proven track record in providing exceptional customer service. He began his career in R&D as a software engineer, and team manager.
  • Stefan Bergstein is chief architect for HP’s Operations Management & Systems Monitoring products, which are part HP’s business service management solution. His special research interests include virtualization, cloud and software as a service.
Follow Us


HP Blog

HP Software Solutions Blog

Labels
The opinions expressed above are the personal opinions of the authors, not of HP. By using this site, you accept the Terms of Use and Rules of Participation