How to use BIG data in my module?

Igor Sysoev igor at sysoev.ru
Fri May 6 08:44:10 MSD 2011


On Fri, May 06, 2011 at 12:38:55PM +0800, XueHJ wrote:

> ----- Original Message ----- 
> From: "Piotr Sikora" <piotr.sikora at frickle.com>
> To: <nginx-devel at nginx.org>
> Sent: Thursday, May 05, 2011 8:26 PM
> Subject: Re: How to use BIG data in my module?
> 
> 
> > Hi,
> > 
> >> in which to load into memory a BIG data (about 800MB),
> > 
> > 800MB isn't really "big data" [0].
> > 
> >> the processing of each HTTP request must use the data,
> For me it really is a large data, huh, huh
> > 
> > Each request must use all or just parts of the data?
> Yes, just parts of the data.
> > 
> >> how do I store the data? What structure to use?
> > 
> > That of course depends on the data and the processing.
> > There is no general answer.
> The reason why the data is loaded into memory, hoping to respond to requests faster.
> But how to achieve it? Use ngx_shared_memory?Asked me a suggestion.

If this is read only data, you can read at at configuration phase
and they will be shared among all workers due to fork().
Also, you can just mmap() the file.


-- 
Igor Sysoev



More information about the nginx-devel mailing list