persistent data : common/best practices

[Jérôme Renard <jr...@ez.no>]

Good evening list,

I am working on a module which interacts with a database by using mod_dbd.

My use case is actually the following :

- I have to fetch a few rows in a specific table (around 10 rows)
   but I do not want to fetch these rows for every request as I know
   they will not change often (but I can not use a configuration file here).

I thought about "caching" the result of this fetch so it is created once
and every subsequent request could use it without hammering the database.

The problem is that I have no clue on how to do it. I mean I can manage with
the code and APR/APR-utils, but how could I do it right ?

Any hint/comments/ideas are much welcome.

Thanks in advance for any help and have a nice week-end.

Best Regards.

-- 
Jérôme :)

Re: persistent data : common/best practices

[Jérôme Renard <jr...@ez.no>]

Hi Joe,

joe@joe-lewis.com a écrit :
>> My use case is actually the following :
>>
>> - I have to fetch a few rows in a specific table (around 10 rows)
>>    but I do not want to fetch these rows for every request as I know
>>    they will not change often (but I can not use a configuration file
>> here).
>>
>> I thought about "caching" the result of this fetch so it is created once
>> and every subsequent request could use it without hammering the database.
> 
> My suggestions :
> 
>   - Populate an apr_table_t with the data.
>   - Use a shared memory segment so that if you use the prefork MPM, each
> process can still access the data (use mutexes to keep things sane).
>   - Use a time stamp on the data entries just so that (after they
> "expire") you can go back and reload the data from the database.
>   - Use a configuration option to set the "expire" value so that the
> module can be adjusted as the load on the db increases.
> 
> I don't have any example code with me right now due to where I am, but you
> should be able to google for it, or even check the apache source for
> examples.  There is much out there.

Thanks for your help! :)

-- 
Jérôme :)

Re: persistent data : common/best practices

[jo...@joe-lewis.com]

> My use case is actually the following :
>
> - I have to fetch a few rows in a specific table (around 10 rows)
>    but I do not want to fetch these rows for every request as I know
>    they will not change often (but I can not use a configuration file
> here).
>
> I thought about "caching" the result of this fetch so it is created once
> and every subsequent request could use it without hammering the database.

My suggestions :

  - Populate an apr_table_t with the data.
  - Use a shared memory segment so that if you use the prefork MPM, each
process can still access the data (use mutexes to keep things sane).
  - Use a time stamp on the data entries just so that (after they
"expire") you can go back and reload the data from the database.
  - Use a configuration option to set the "expire" value so that the
module can be adjusted as the load on the db increases.

I don't have any example code with me right now due to where I am, but you
should be able to google for it, or even check the apache source for
examples.  There is much out there.

Best wishes!
Joe