Think about the data schema!
use poi;
doc = {
'name' : 'A440',
'tags' : {
'highway' : 'secondary',
'oneway' : 'yes'
}
};
db.poi.createIndex( { 'name' : 1 } );
db.poi.createIndex( { 'tags.highway' : 1 } );
db.poi.createIndex( { 'tags.oneway' : 1 } );
// Road with name=Strand
db.poi.find( { 'name' : 'Strand' } );
// All roads
db.poi.find( { 'tags.highway' : { '$exists' : 1 } } );
Lots of indexes are required
Easy to extract data from
or:
doc = {
'tags' : {
{ 'k' : 'name', 'v' : 'A440' },
{ 'k' : 'highway', 'v' : 'secondary' },
{ 'k' : 'oneway', 'v' : 'yes' }
}
);
db.poi.createIndex( { 'tags.v' : 1, 'tags.k' : 1 } );
// Road with name=Strand
db.poi.find( {
'tags' : { '$elemMatch' : { 'k' : 'name', 'v' : 'Strand' } }
} );
// All roads
db.poi.find( { 'tags.k' : 'highway' } );
One index required
Good for finding key/value combinations
Extra index needed for "all roads" question (on %tags_indexed.k%)
or:
doc = {
'tags' : {
'name=A440',
'highway=secondary',
'oneway=yes'
}
);
db.poi.createIndex( { 'tags' : 1 } );
// Road with name=Strand
db.poi.find( { 'tags' : 'name=Strand' } );
// All roads
db.poi.find( { 'tags' : /^highway=/ } );
One index required
Good for finding key/value combinations
Good enough for doing the "all roads" question