Customer Success with Effectual’s Probe & UCMDB Performance Tuning

Customer Success with Effectual’s Probe & UCMDB Performance Tuning

 

At Effectual, universal success for HPE UCMDB customers is our vision. That’s why we share our revolutionary MySQL Probe Tuning findings for users on older versions, and our UCMDB Performance Tunings that affect all users, regardless of version. When we published our Performance Tunings, we asked customers to share their own results with us and with HPE. That’s why we offer our Packaged Integration Enhancement (PIE) solutions and services that streamline your road to success with UCMDB & Universal Discovery. We recently received the following testimonial from IT applications analyst Mike L., the CMS Subject Matter Expert for a major U.S. oil & natural gas corporation, who had great results with our tuning suggestions, and just a few hours of our support, from his nearly 5 year-old UCMDB project:

My name is Mike L. and I support HPE UCMDB at [Corporation]. I’m writing to inform you of the success I’ve had implementing the performance tuning recommendations as set forth by Effectual Systems Group, Inc.

When I inherited our CMDB, the Discovery jobs were using out-of-the-box TQLs as trigger inputs. Our 2 Data Flow Probes were severely overloaded, attempting to scan about half of a Class A network.

Prior to tuning the probes, we were faced with an extremely slow job completion rate. We were forced to look at our total discovery as a week-long process. The only option set before us was to add more hardware. The jobs would run for days and oftentimes would not complete in time for the next scheduled run.

Our largest jobs were continually competing for resources and taking more than 10 days on average to complete a full run. On several occasions the stuck thread count was exceeded and the DFP would either restart itself or need to be restarted from running out of free memory. Shortly before contacting Effectual, I got the jobs scheduled out where they wouldn’t overlap, but I still needed to give the Host Connection and Resource jobs 3 to 4 days to complete.

At this time I had approximately 65,000 Node CIs. A vast number of these were either unqualified, not fully discovered, or duplicates. I began working with Effectual’s support to tackle this problem by modifying the TQLs in the out of the box (OOTB) Discovery jobs.

Mike shared his results of this first round of TQL tuning on the OOTB packages, which averages out to a 195% performance improvement.

The primary use case for us is to identify all active servers in our worldwide environment. The improvements to the OOTB discovery jobs have practically eliminated duplicate server CIs. This had been a major problem when it came to getting an accurate server count.

Alongside this issue was a backlogged Aging mechanism, which thanks to proper discovery is now no longer a problem. Since then, the number of actual nodes has dropped from 65,000 to around 14k. The duplicates were allowed to age out because we weren’t constantly “touching” the bad CIs in the UCMDB with each Discovery run. The time we save from dispensing with the need to manually “massage the data” has allowed us a much faster turnaround. We now have the capability to deliver information as soon as it’s needed.

Socializing UCMDB as the best source for accurate and timely configuration data has been our plan all along. However the knowledge gap between what comes OOTB and what actually works has prevented us from accomplishing this. It’s comforting to know that professionals such as Effectual exist and have overcome these challenges, paving the way for UCMDB to become the “Source of Truth” in our environment. Once this reliability and completeness has been established, impact analysis will be our next big step. The number of discovered servers continues to rise, so our customer has renewed faith that UCMDB can take over as the source of truth.

Shortly thereafter, Effectual also shared their procedures and MySQL Probe Tuning Release with me and I applied it to 1 of my 2 probes so I could have a side-by-side comparison. Actually, the probe I applied tuning to was the more heavily utilized of my probes. Since the tuning I’ve continued to add discovery ranges to the tuned probe; it continues to outperform the un-tuned probe. The time saved by tuning our probes gives us greater confidence in our data since discovery can now run repeatedly throughout the week, while at the same time cutting our hardware costs in half.

Mike also shared his results of the MySQL Probe tuning changes before and after. Combined with the TQL tunings, he experienced a 556% performance improvement over OOTB.

Since having this success in my Production environment, I have also applied these improvements and the additional Application Server improvements to my test environment. My preliminary testing indicates that with all of the combined configurations I’m reducing the time spent on Host Connection jobs by 60% to 80% [on top of the first two rounds of  tuning] (this is starting with the modified discovery jobs, so the change is not as drastic as when I started with the OOTB jobs).

After implementing the UCMDB and MySQL Dataflow Probe application tuning changes we recommended, Mike experienced a further 450% improvement, on top of tuning the Discovery TQLs and following our MySQL Probe Tuning guide. We’ve graphed some of his improved Discovery job times below to illustrate the positive impact of tuning the way UCMDB processes data (Note that Before times include the already greatly increased performance of the first two rounds of tuning).

We also had a very problematic integration with Service Manager. Through recommended tweaks on both the UCMDB and SM side, the integration is now running smoothly and to satisfaction of our internal customers. It is very exciting to know that Effectual has leveraged 3rd party systems for integration, thus saving license costs for their customers. We did not know this was a possibility, much less be able to do it on our own. We will certainly be working with Effectual to expand our integrations in the future.

In conclusion, I am extremely thankful to Effectual for the hard work they’ve done in pinpointing, testing and documenting these configuration improvements and I would recommend them to anyone! Secondly, as an HPE customer I strongly recommend that these improvements be integrated into the out of the box HPE UCMDB solution. Should anyone from have any questions for me, I’d be pleased to answer them.

Mike L.

With a mere 5 hours of Effectual support, Mike was able to increase his enterprise’s overall UCMDB performance by an astonishing 1201%!


At Effectual we truly believe in HPE Software’s value and the synergy of their portfolio. We’ve focused on simplifying the challenge of making integrated HPE solutions an enterprise reality while pursuing advanced integration use cases such as SACM. We’ve achieved that goal by making the most complete integration packages possible, for each direction, between each tool, streamlining the path to data quality. Our philosophy is based on out of the box compatibility, our vision is for every HPE ITOM customer to get the kind of results Mike has experienced at his enterprise.

To ask about our solution packages, or for information to contact Mike directly about his experiences, don’t hesitate to

Contact Us

 

To experience similar fast results to those Mike has seen, check out our UCMDB Performance Tunings!

Download Technical White Paper

 

And our MySQL Probe Tuning Package!

Download Probe Tuning Package