seanpringle / n3

n3

A nested map of unsigned 64-bit integers.

[ ID => { key: value } ]

map <uint64_t, map <uint64_t, uint64_t>>

• No restriction on what each number means:
  • IDs might be UNIX timestamps / keys might be metrics / values might be scaled fixed-point measurements
  • IDs might be object identifiers / keys might be references to other objects / values might be weights
  • IDs might be RDBMS surrogate keys / keys might be table fields / values ... etc
• Numbers can have text aliases for query readbility
• IDs are sorted in ascending order

Example: Time Series

alias 1 cpu
alias 2 memory
alias 3 disk

insert 1454376744 cpu 39 memory 765504 disk 6245996
insert 1454376864 cpu 95 memory 781636 disk 6246356
insert 1454376984 cpu 51 memory 777996 disk 6246756
insert 1454377104 cpu 33 memory 794044 disk 6247084
insert 1454377224 cpu 34 memory 797196 disk 6247464
insert 1454377344 cpu 38 memory 808208 disk 6247880
insert 1454377464 cpu 33 memory 825500 disk 6248244
insert 1454377584 cpu 98 memory 828884 disk 6248616
insert 1454377704 cpu 42 memory 829188 disk 6248936
insert 1454377824 cpu 27 memory 839720 disk 6249332

select mean(cpu) as cpu_pct max(memory) as mem_kb max(disk) as disk_kb from 1454376744:1454377824:300

0 4
1454376744 cpu_pct 61 mem_kb 781636 disk_kb 6246756
1454377044 cpu_pct 33 mem_kb 797196 disk_kb 6247464
1454377344 cpu_pct 56 mem_kb 828884 disk_kb 6248616
1454377644 cpu_pct 34 mem_kb 839720 disk_kb 6249332

Query Syntax

Insert

insert id key val [ ... keyN valN]

Delete keys (preserving IDs) by range

delete key ... keyN from id:idN

Delete IDs and keys

delete id:idN

Select keys by range

select key [ ... keyN ] from id:idN

Select keys by range with aggregation

select func(key) [ ... funcN(keyN) ] from id:idN:step

Select keys by range with aggregation, no gaps

select func(key) [ ... funcN(keyN) ] from id:idN:step fill

Supported Functions

• min
• max
• mean
• median
• first
• last
• sum
• diff