Tuesday, November 25, 2014

Total Cost of Ownership of BPPM - Part 7: Best practice - Name convention

Many things you can do in BPPM implementation to lower the total cost of ownership of BPPM.  Some of them are big such as architecture design.  And some of them seem to be small such as name convention.  However small they seem to be, they help minimize 'Winchester House' syndrome and lower the total cost of ownership of BPPM.

I would like to start best practice discussion with name convention.  BPPM doesn't post any requirement on how you name your BPPM components including files, cells, integration services, clusters, CMA tags, etc.  If there is no name convention in place, each person would end up naming BPPM components any way he/she likes.  Sooner or later, your BPPM implementation would become a Winchester House. 

Having a name convention in place at the beginning of BPPM implementation is a small effort but it pays many times back after the implementation is completed.  The name convention needs to be enforced at all time during the implementation and after the implementation.  Trying to rename things afterwards is a painful and error-prone process.

Pick up a 2 or 3-letter prefix for your organization.  It will help distinguish your custom files from BPPM out-of-box files.  For example, I used prefix C1 for CapitalOne and prefix CTI for CitiMortgage. Use this prefix for your custom PATROL KM files, PSL files, cell knowledge base MRL files, BAROC files, shell scripts, perl scripts, batch files, JAVA scripts, etc.

For custom PATROL KM, also pick up a short name for your KM.  I also prefer to use all capitals for KM short names.  Include this short name in all file names related to the KM.  This will make your KM easy to package and deploy without missing any file.  If the KM requires specific pconfig file, I would use the same name convention for the pconfig file.  For example, here are the file names I used for a custom CACHE KM: XYZ_CACHE_main.km, XYZ_CACHE_db.km, XYZ_CACHE_db_collector.psl.  XYZ_CACHE.cfg.  XYZ here is the organization prefix. 

For cell names, never ever reuse the same name in the entire organization.  Since cell names are prompted during BPPM installation, you need to have a discussion and decide on their names before you start installation.  The information needs to be included in cell names include: 1) Environment: dev, QA, or production. 2) Event source: PATROL, SCOM, external, correlation, server, etc. 3) Type: H/A or standalone.  For example, I named the cell located on the first clustered PATROL integration service in dev environment as Dev_PATROL1_HA. 

CMA tags also need to be determined in advance.  It can usually be divided into two different ways: by infrastructure (OS, DB, application) and by user environment. For example, you may have CMA tags like Windows_Base, Oracle_Canada, Peoplesoft_HR, etc.  The less overlap between tags, the easier to use and maintain them.  Be careful when making changes to existing CMA tags after implementation is completed because all PATROL agents with the same tag names will have their pconfig updated automatically. 

No comments:

Post a Comment