Sizing baselines

This topic describes the deployment architecture for the IT Service Management (ITSM) initiative. The combined architecture for the IT Service Management value paths is presented with the default security configuration, showing the flow of information between individual products. Because BMC Atrium CMDB is hosted in the BMC Remedy Action Request System server (BMC Remedy AR System server), it is represented in the BMC Remedy AR System environment. The architecture does not include high availability, but guidelines for building a high-availability environment are provided in High availability.

Several core technologies must be considered when sizing an IT Service Management environment. Sizing of each product in the value path is based on the individual load on each product's sample use case, and several combined use cases and value paths were tested for the Performance benchmarks.

Environment baseline guidelines

The following table refers to both concurrent users and managed devices. Use this information to size your environment.

Use the following definitions to help understand this task:

Concurrent users is used to estimate the number of simultaneous connections to the mid tier or BMC Remedy AR System servers. If your organization is using a significant number of web-based clients (utilizing the mid tier) and the thick-based BMC Remedy client tool, you should calculate the totals for each and use the total number as the estimate for concurrent users.
The term managed device has historically referred to devices that have an IP address and can therefore be remotely connected to over an Ethernet network. However, this might not always be the case. It might be beneficial to track other items as well. BMC Atrium CMDB enables you to track items that do not have an IP address, such as telecommunication network links (T-1s, DS3s, and so on) or desk telephones that can be tracked another way, such as through MAC addresses or serial numbers.

Hardware considerations for sizing

The numbers in the following table are based on performance lab benchmark tests and examples observed from consulting services and customer experience.

Hardware sizing for IT Service Management components

Last modified by Bruce Cane on Jun 18, 2013

reference deployment best_practices sizing hardware_recommendations database_sizing database_tier database_tier_sizing

Comments

Vinod Gaidhani

It is very difficult to understand if the baseline given above is for Virtual or Physical platform. Is it mentioned somewhere else?

Apr 22, 2014 02:49
1. Bruce Cane
  
  Thanks for your question. I'll pass it along to the person who owns this information to get an answer.
  
  Apr 22, 2014 10:39
1. Bruce Cane
  Hello Vinod:
  Here is a response to your question. It comes from one of the people who drafted this information originally:
  The sizing guidelines are independent of whether the environment is physical or virtual, i.e., it applies to both. Some additional points to consider:
  The CPU sizing is based on physical cores and should NOT include logical threads (hyperthreading).
  For virtual environments, consider mapping a vCPU to a physical core. Now follow the sizing guidelines.
  For virtual environments where performance/scalability is critical (i..e. production environment), do not oversubscribe.
  Apr 23, 2014 09:21
Vinod Gaidhani

Perfect answer Bruce!!!
Thanks a lot for quick response.

Apr 23, 2014 10:00
1. Bruce Cane
  
  Good! Thanks Vinod, I'll pass this on to our subject expert. Good luck.
  
  Apr 23, 2014 11:40
Mark Twomey

Are there any recommendations for IOPS requirements for small, medium and large ITSM deployments for green field sites where SAN is required? This is covered alot in other BMC products so would be useful for ITSM

Aug 06, 2014 07:25
1. Catherine Siderine
  
  Hi Mark,
  Thanks for your question. I'm checking with a subject expert on this.
  Regards,
  Cathy
  
  Aug 06, 2014 06:01
1. Bruce Cane
  
  Hi Mark,
  I have spoken to the deployment architecture team about this. Some information was provided for the 8.0 release, which is equally applicable to 8.1. You can find a reference to read/write operations per section for large implementations at the bottom of this page:
  https://docs.bmc.com/docs/display/public/itsm80/Online+test+and+results
  The deployment architecture team let me know that they will make IOPS part of their standard reporting going forward.
  Thanks for the question.
  -Bruce.
  
  Jan 19, 2015 01:47
Mark Twomey

Thanks Bruce
I cant seem to find any reference to iOPS. Can you send me the excerpt?
Mark

Jan 20, 2015 03:08
1. Bruce Cane
  Mark,
  My apologies, I provided you with the wrong link.
  This is the correct link (I also updated my original response, above): https://docs.bmc.com/docs/display/public/itsm80/Online+test+and+results
  You'll find this is at the very bottom of that page:
  Following is the I/O footprint for the online tests:
  Peak network bandwidth from each database node to the storage array = 45 MB
  Average physical database reads/writes per second per node = 1025/80
  The deployment and sizing people pointed to the second bullet in reference to IOPs.
  Hope this helps.
  Jan 20, 2015 08:47
Mark Twomey

Thanks Bruce
Another point I would like to add from alot of experience in deploying ITSM is around the database sizing. The sizes put forward for both medium and large in my experience are too low. I really think the database sizing needs to take into account volumes per time period and size should be aligned to that. Using concurrent users to estimate database size isn't really ideal and should also take into account integrations and database processing required for this. I know of companies who would theoretically fit into small or medium who have quadruple the size of database. Any thoughts on this?

Mark

Feb 06, 2015 08:44

Log in or register to comment.

Sizing and deployment considerations

Sizing diagrams