[Rd] how useful could be a fast and embedded database for the R community?

Dirk Eddelbuettel edd at debian.org
Wed Dec 24 20:22:54 CET 2014


On 24 December 2014 at 10:49, joanv wrote:
| I've already done some benchmarks, again leveldb and sqlite, are quite slow
| compared with my release.
| 
| They cannot be used very intensively in a huge computation, because the low
| performance.
| 
| Yes, billions ( American ones : thousands of millions) with a generous RAM.
| 
| More details at: www.vulcandb.com
| 
| My main concern, is where can feet better such a database. In what fields,
| or in what kind of calculations.

I second what Elijah said:  "working code". 

So far I see just a (very pretty) website, but stricly no code. Not good.

I work in a industry where we 
  a) do use billions of time series points and 
  b) are latency and performance sensitive

We like flat (binary) files as well as mmap a lot, and do a lot of C++ for
performance where we'd never ever dream about embedding R.  But we do course
use R for analysis and then embed quite some C++ into to "do stuff".  We use
Redis (out of process) for a few things, but obviously not "raw data". I like
what I see from influxdb.com, but we are now off-topic for this list.

So colour me "interested" -- but please show some code, or people will tune
out pretty quickly.

Dirk

-- 
http://dirk.eddelbuettel.com | @eddelbuettel | edd at debian.org



More information about the R-devel mailing list