Load Multiple HDBs in one KDB Process

2025.08.16

With par.txt and .Q.bv[] you can load discrete HDB directories into a single KDB process.

For this to work, we assume that:

• Tables have unique names across the discrete databases.
• Symfiles are unique i.e. cannot have 2+ symfiles named sym for the different databases.
• Symfiles are not stored at the root of discrete databases

Example setup:

# discrete hdb
internal
├── 2025.08.14
│   └── positions
│   └── orders
│   └── executions
├── 2025.08.15
│   └── positions
│   └── orders
│   └── executions
mktdata
├── 2025.08.15
│   └── minute_bars
│   └── daily_bars
symfiles
└── sym_internal
└── sym_mktdata

The unified hdb contains the par.txt that lists all the discrete hdbs you want to load in one process. The database also includes symbolic links to the sym files.

# unified hdb
db
├── par.txt
├── sym_internal -> ../symfiles/sym_internal
└── sym_mktdata -> ../symfiles/sym_mktdata

# cat db/par.txt
/home/jp/kdb/internal
/home/jp/kdb/mktdata

To load the database and run .Q.bv[]

$ q db
q)tables[]
`minute_bars`daily_bars
q).Q.bv[]    / will build vp
q)tables[]   
`positions`orders`executions`minute_bars`daily_bars