Node structure statistics

[Jukka Zitting <ju...@gmail.com>]

Hi,

Here's some statistical information about common node structures as
included in a vanilla CQ5.5 installation.

Number of child nodes:

    0 child nodes = 73%
    1 child node = 20%
    > 1 child nodes = 7%

Number of properties:

    0 properties =  0% (all nodes have at least jcr:primaryType)
    1 properties = 5%
    2 properties = 2%
    3 properties = 20%
    4 properties = 5%
    5 properties = 40%
    6 properties = 22%
    7 properties = 1%
    8 properties = 1%
    9 properties = 1%
    > 9 properties = 3% (all nodes have less than 200 properties)

The same sets of property names (and types) repeat a lot - the top ten
such "effective node types" accounted for 82% of all nodes.

Number of properties of different types:

    STRING = 46% (45% single, 1% multi)
    NAME = 31%  (20% single, 11% multi)
    DATE = 17% (17% single, 0% multi)
    BINARY = 2% (2% single, 0% multi)
    LONG = 2% (2% single, 0% multi)
    other = 2%

These numbers should help us design more efficient data structures for
tracking nodes. More on that later.

BR,

Jukka Zitting