Naming convention for Database services at CERN

Last updated: 27.October.2005

Having a general naming convention across all database services at CERN will help not only account management for administrators but also allow experiments and users to prepare their applications for the several stages of developing.

With the starting principle that each developer uses at most one account per project (but one project can have more than one development account) we would propose the following naming convention:

Project name

The project name should be less than 15 character long (for easiness) prefixed by the experiment name followed by an underscore (ALICE_, ATLAS_, CMS_, LHCB_, LCG_, NOLHC_). Examples of project name are: LHCB_BOOKKEEPING, CMS_TRANSFERMNGM, ATLAS_PIXEL, ALICE_TRACKER, CMS_MUONDT.

Usernames

The database usernames should start by the project name and having as suffix:

For integration and production:

• • none for human account with schema and data
    • assigned with: CERN_DEV_ROLE and CERN_DEV_PROFILE

• • Option 1:
  • _W for application account with write access to necessary tables/views/etc
  • _R for application account only with read access
    • assigned with: CERN_APP_ROLE and CERN_APP_PROFILE

• • Option 2:
  • _APP for more complex access control, to use in conjunction with database roles
    • assigned with: CERN_APP_ROLE and CERN_APP_PROFILE
  • Examples: CMS_TRACKER, CMS_TRACKER_W, CMS_TRACKER_APP

For development:

• none when there is only one account needed/requested
• _<NICE> when more than one is needed for same project, where <NICE> is the CERN NICE login of the developer
• Examples: ATLAS_MUONDT, ATLAS_MUONDT_MANJO

Roles

When needed a fine grain control, the owner can create roles with certain priveliges. The application then need to enable the roles (with or without password) after connecting to the database. The role names should start with the project name followed by ‘_’:

• Examples: ALICE_ELECTRONICS_BK_UPDATER

Tablespaces

The user should never specify the tablespace name in their code or object creation. It should always use the assigned default tablespace except in very particular cases. The tablespaces names that will be used are described below:

For integration and production:

• Because of easiness for user import/export, it is better to use tablespaces per project. The name should start by the project name and having as suffix:
• _DATA01 for tables/indexes/views
• Others when data expected bigger than 1TB or if required for BLOBS
• Example: ALICE_PIXELDT_DATA01 (for first tablespace of the user)

For development:

• Use common tablespace for every project in the database
• DATA01 data

Connection string

On production level every project should have one connection entry in tnsnames.ora for easy management. This entry should be the same as the project name. Example is:

ATLAS_MUONDT  =
  (DESCRIPTION=
    (ADDRESS=(PROTOCOL=TCP)(HOST=rac11-v.cern.ch)(PORT=1521))
    (ADDRESS=(PROTOCOL=TCP)(HOST=rac12-v.cern.ch)(PORT=1521))
    (CONNECT_DATA=(SERVICE_NAME=atlas_muondt.cern.ch)))

Note: Applications should be ready to change string when requested