Developers of Sympa

[IKEDA Soji <address@concealed>]

On Fri, 25 Jan 2013 09:31:34 +0100
Marc Chantreux <address@concealed> wrote:

> On Fri, Jan 25, 2013 at 02:14:32PM +0900, IKEDA Soji wrote:
> > To cross over process boundary, I suppose other mechanisms such
> > as shread memory (e.g. Cache::FastMMap) would be used by 6.3.
> 
> IMHO, the goal would be
> * remove the DBMS: we really don't need it and it's anoying for
>   administration (for example, i'm really sad i can't use git
>   to store the history)
> * add a cache system which could be networkable because really huge
>   sites could need a distributed infrastructure (i would choose redis)
> * use robot and listname as prefix to gain the expected granuarity 
> * don't check the state of the configuration file in the FS: use
>   temporary files and "push to cache" mechanism instead.
> rewrite the parser to speed the things up: a site with less than 500
> lists (i think this is the majority of the sites) would run fast enougth
> without any cache system.

It's ambitious idea.  During work on db_list_cache branch (now
merged into 6.2 branch), I in fact daydreamed Sympa on Hadoop,
MongoDB etc.:-)  There may be slightly more works to them be true,
though...

Current Sympa evidently provides vertical scalability (more
memories, more cores, ...).  Occasional questions about "clustering
Sympa" on sympa-users list suggest needs of horizontal scalability.

IMHO "Distributed Sympa" you supports will give advanced scalability. 


> regards
> marc 

Regards,

--- Soji

-- 
株式会社 コンバージョン  セキュリティ&OSSソリューション部   池田荘児
〒231-0004　神奈川県横浜市中区元浜町3-21-2 ヘリオス関内ビル7F
e-mail address@concealed  TEL 045-640-3550
http://www.conversion.co.jp/