Part 1: More Than Just A Tuneup
We are thrilled to announce a new era for HPE UCMDB and UD customers! Effectual has recently completed an extensive series of tuning and optimizations against the MySQL version of the HPE Data Flow Probe with these amazing results:
- Probe performance improves a minimum of 7 times to a peak average of 366 times better throughput using the same resources. Yes. Seriously.
- Probe survivability and reliability can be increased to 100%
- The troublesome “Clear Probe Cache” will no longer be a regular feature of Probe use
Since it is important to us that you learn more about Effectual’s services and solutions, and our dedication to the best possible customer outcome with HPE Software, we have provided a complete probe tuning guide free for your use.
These are changes you must make to your HPE UCMDB Probes (version 8, 9 or 10.01) that require only a few hours of investment to save you thousands.
We Did the Work So You Don’t Have To
The Probe (as it is commonly referred to) is not just a process sitting on a box, but a whole ecosystem. And yet it has been completely ignored. We ignored it too. Until recently.
So we did some testing. Then some tuning. Then a lot more testing. We observed, learned more about MySQL and made a little change, then tested each change. Over 700 times.
And we think you’ll find the results surprising. In fact, we think they will revolutionize the entire concept of the Probe as we know it. And since the shortcomings of the Probe are often misinterpreted as shortcomings of HPE UCMDB itself, our findings may just change your understanding of the use cases, and enterprise business value, of the entire application.
Over the course of a this two part blog series, we will be sharing our findings and explaining in detail how they affect you. There are two important things we want you to take away from these blogs: That Effectual will be an incredible asset for your organization and, more importantly, that you absolutely must make these changes to your Probes.
Are Misconfigured Probes Really That Big of a Deal?
In a word, yes. If you’re an HPE UCMDB, BSM/RTSM or UD customer using a MySQL version of the Data Flow Probe, the moment your Probe database passes a certain size these problems start, compound, and get worse over time. The threshold for these Probe performance issues is surprisingly low. This is never a question of if, but rather just a question of when. For many HPE UCMDB customers the issues below aren’t a list of problems that might happen, but rather problems that they have lived with for several years.
Above: Just the query to determine the database size processes more than three times greater data, more than sixty two times faster after tuning.
Without Effectual’s tuning and changes:
- Your discovery jobs run too long and often fail
- Your federation jobs run too long and often fail
- Your integration jobs run too long and often fail
- Large jobs will never complete without constant restarts of the probe
- You make liberal use of “Clear Probe Cache,” a two steps forward, 1.8 steps backward function, which compels you to repeat the same work over and over again
- Your Probe can enter a state (often once or more daily) where it seems to “hang,” but has actually stopped functioning, and needs to be restarted
- Probe problems prevent the collection of parts of data, which can age out, causing a whole range of issues with relationships and models, and gaps in your HPE UCMDB
- Your Probe result data can become corrupted due to OS level swapping, returning bad results. This causes excess overhead in the CMDB reconciliation process, or can even create problem data that sneaks past into your HPE UCMDB
- Because of the above scenarios, you end up with many more Probes (10-20+), and many more problematic environments to maintain
- Your organization will wrongly perceive the HPE UCMDB itself as a “slow” product, limiting your political and technical capability to do more integration, as well as your ability to increase value by bringing in more source data
- You will have increased opportunity costs, diminished confidence, and less support for the UCMDB solution from both business and technical stakeholders
All of these problems are removed with just a few simple changes to the Data Flow Probe, the mechanical logging change, and proper sizing of the memory and available resources. The impact is that great. While these changes may appear simple – and essentially they are – the benefit and positive impact on your business and HPE Software experience is gigantic.
Our (Actual) Results
Pictured below are the results of two of our production integration jobs. Both sets show that the previous (un-tuned) fastest version of the job is far slower than our current (tuned) slowest job.
These are actually really large and complex jobs. So why share only these big jobs? Our normal, everyday jobs (such as those moving IT Process Records from Service Manager to UCMDB now take less than 5 seconds from start to stop. These regular-sized jobs are now so fast that they don’t even show up in the UCMDB data access layer log.
While looking at the results above, keep in mind that no additional resources were added for the purpose of these outcomes. Only database changes were included. These larger jobs perform even better with additional server memory and processing resources.
See For Yourself
Our completely free Effectual Probe Tuning Package (download below) includes: A detailed guide to removing the mechanical problem from your Probe, a tuning section, a replacement my.ini configuration file, and SQL queries you can run to maintain the best possible discovery or integration/federation MySQL Probe available.
Achieve a Better HPE UCMDB Outcome
Effectual Systems Group is an HPE partner that provides automation and solutions for HPE UCMDB, and the entire HPE ITOM Service Management tool suite and specializes in creating our Packaged Integration Enhancement (PIE) solutions for some of the world’s largest, savviest, most complex IT enterprises. Ultimately, it is our PIE customers that will benefit most from Effectual’s optimizations of the Probe database.
Our PIE solutions and framework can instantly save you thousands of hours of development and consulting. To learn more about PIE:
or click below to
What are the specific reasons why Probe performance was so poor for so long? Why didn’t anybody notice until now? Why Effectual? How can the recommended changes actually resolve all of the serious problems and pain points you’ve experienced with the Probe?