Air traffic queries in InfiniDB: early alpha

As Calpont announced availability of InfiniDB I surely couldn’t miss a chance to compare it with previously tested databases in the same environment.
See my previous posts on this topic:
Analyzing air traffic performance with InfoBright and MonetDB
Air traffic queries in LucidDB

I could not run all queries against InfiniDB and I met some hiccups during my experiment, so it was less plain experience than with other databases.

So let’s go by the same steps:

Load data

InfiniDB supports MySQL’s

statement and it’s own

utilities. As

is more familiar for me, I started with that, however after issuing LOAD DATA on 180MB file ( for 1989 year, 1st month) very soon it caused extensive swapping (my box has 4GB of RAM) and statement failed with


was more successful, however it has less flexibility in syntax than

, so I had to transform the input files into a format that

could understand.

Total load time was 9747 sec or 2.7h (not counting time spent on files transformation)

I put summary data into on load data time, datasize and query time to Google Spreadsheet so you can easy compare with previous results. There are different sheets for queries, datasize and time of load.


Size of database after loading is another confusing point. InfiniDB data directory has complex structure like

so it’s hard to day what files are related to table. But after load, the size of 000.dir is 114G, which is as twice big as original data files. SHOW TABLE STATUS does not really help there, it shows

with totally misleading information.

So I put 114GB as size of data after load, until someone points me how to get real size, and also explains what takes so much space.


First count start query

took 2.67 sec, which shows that InfiniDB does not store counter of records, however calculates it pretty fast.


Another bumper, on this query InfiniDB complains

Ok, so InfiniDB does not support DERIVED TABLES, which is big limitation from my point of view.
As workaround I tried to create temporary table, but got another error:

As warning suggests I turned

, which is: