DX NetOps

 View Only

Introducing CA Performance Management r2.3.4 and helpful tips from our Early Adopters Program

By Dan_Holmes posted Sep 22, 2014 10:51 AM

  

 

Our Team is very excited to be introducing a new version of CA Performance Management this week. This release contains some great new architecture and feature enhancements.

 

Over the past 3 weeks we have been working closely with a subset of our customers deploying CA Performance Management r2.3.4 as part of a new Early Adopters program. We have been piloting this program with really great success – members of our SWAT, Engineering, & Support teams work closely with customers to observe the total customer experience upgrading or installing CA Performance Management 2.3.4 and CA Mediation Manager 2.2.6. In the future I hope you all have a chance to participate.

 

The purpose of this program is to improve the way our R&D teams observe and to learn what it’s like for our customers  to deploy our software so that we can not only improve the product but we can improve our processes and supporting utilities, documentation, training programs, etc.

 

I was talking with our team this week and we thought it might be nice to share with the community some of the things we learned to help you be successful upgrading or deploying this release.

 

  1. Vertica enforces non-LVM deployment:CA Performance Management has been architected to use a ‘big data’ type of data repository and with that comes some new challenges. We upgraded support for Vertica 7 to improve stability, performance, ease of deployment, and access to better administration tools. With this upgrade we learned that Vertica now enforces deployment of their software in Non-LVM environments only. This motivated a lot of research by our team and conversations with Vertica to understand the rationale. We learned that the risk to data corruption is much too high for reasons I won’t go into detail here but in summary we agree that we must only support deployment of our product and any backups of our data repository on non-LVM file systems.

     

    Chances are you will run into this. LVM is a default configuration and a lot of system admins like the capability so as we learned during customer validation and early access many of you will discover that when you go to install or upgrade you will get a failure during install that you will need to overcome.  The benefit of our customer validation and early access program is that we have taken the time to understand and provide training to our support, services, and presales teams as well as provide you with good documentation in our release notes. PLEASE take the time to review the Release Notes (Sec 3 and Sec 6.1) with your system admins. We strongly believe that data corruption would be a very expensive failure to overcome with a deployment of CA Performance Management and as a result we strongly recommend that you follow our installation and upgrade requirements to ensure you have a good experience with our solution.

     

  2. Early Access Customers did find a couple mistakes:  During Early Access we uncovered a couple defects and have solutions already on deck.
    1. When you do a fresh install of 2.3.4 RTM kits or do an upgrade to 2.3.4 RTM kits, if you discover Huawei device that supports both “Huawei CPU” and “Huawei Entity CPU” VCs, you will see CPU components getting discovered with both the VCs “Huawei CPU” and “Huawei Entity CPU”. This causes Average CPU Utilization data on System Health tab of Huawei device to be wrong since Huawei CPU components show aggregate CPU Utilization data of individual Huawei Entity CPU components. This is happening due to the new Vendor Cert Priority Grouping functionality we added in 2.3.4 release.The following tech Tip was added to help you resolve this situation if you see this happening in your environment. In addition we have a solution in our monthly update kit that will take care of it. Tech Tip : PM 2.3.4 - Discovering CPU components for Huawei devices within the upcoming 2.3.4 release

       

    2. The second issue pertains to the CA Mediation Manager Migration Scripts that were shipped with 2.3.4 Early Access. A few issues were uncovered by one of our customers performing a migration of some device packs and we determined that an improvement to the script was needed for this to work successfully. The following tech Tip was provided by support to help you determine if you should use a new script that is included with this bulletin:  Tech Tip : PM 2.3.4 - CAMM 2.2.6 Migration Script improvements

       

       

  3. We choose to deliver our segmentation scripts outside the product with this release and we learned during early access this may be confusing for customers so we have updated the GA letter that is sent announcing the release of 2.3.4 but I thought I would also mention it here. One of the great things with this new architecture is the opportunity to take advantage of horizontal scaling of our Data Repository. As we introduced this support prior to 2.3.3 we were not segmenting the data in the most optimal way and were writing copies of the data to each node. Starting in 2.3.3 we added support for more intelligent distribution of the data to a subset of the nodes in your cluster. What this means is that if you have data from installations prior to 2.3.3 you will want to run these segmentation scripts to organize properly the data according to the new format. The good news is that as you do this you get an instant savings in Disk utilization (30% and higher) and you now have an architecture you can scale out as you want to collect more, do more analytics, or support more users performing concurrent tasks. It is very important you run these scripts and as I shared in the GA letter you can find these by going to support.ca.com and then searching for product CA Performance Management. You will find the script on the right hand side of the page under popular links.

  

In addition to finding and fixing these issues we learned a lot and have some great ideas brewing on how to improve the Time to Value and overall customer experience installing, upgrading and using CA Performance Management.  One of my favorite parts of the Early Access program is the direct interaction with customers and our R&D teams. The feedback on stability and performance has been very positive as well as support for new features like score carding and data collector load balancing.

 

Please continue to share your feedback to the community. Stay tuned for upcoming EOS Demos and access to our customer validation builds. If you would like to be part of our CA Performance Management 2.4 and/or CA Mediation Manager 2.4 Early Access programs send me a note at daniel.holmes@ca.com

 

9 comments
0 views