Prev: How to stop E_DEPRECATED messages in the PHP log?
Next: How to Install Roadsend Compiler on Fedora
From: "Michael A. Peters" on 10 Jul 2009 15:05
workerholic(a)studysite.eu wrote:
> hi andrew i think you understand my problem a little,
> but if 100 user load this query at the same time, the two mysql server
> had a lot to do!
> so i think to cache this query as xml to the application server local
> make thinks faster,
> but, i would like to have the same performance to read this xml document
> as read the query from mysql server...
> i dont know why php is so slow to read the xml file...

Are you saving to file or caching as a query result?
Also note that you can cache an array of rows (at least with APC but I
suspect memcache as well) - say my_fetch(key) is you function to fetch
from cache and my_store(key,data,life) is your function to store.

$result = my_fetch('big_query');
if (! $result) {
$sql = 'your query';
$rs = mysql_query($sql);
while ($row = mysql_fetch_object($rs)) {
$result[] = $row;
}
my_store('big_query',$result,21600);
}

No xml involved and you can loop through the results.

If you'd rather do it as xml, you can cache the xml as a string and then
fetch it, importing it into a DOM or whatever to extract your results.

$xml = my_fetch('queryResultAsXML');
if (! $xml) {
generate xml and cache it
}
$dom = new DOMDocument('1.0','utf-8');
$dom->loadXML($xml);

Not sure what you are doing, apoligize if these suggestions are useless
or already considered.
From: "workerholic on 10 Jul 2009 17:25
first thanks to all who have read ;-)

your solution looks like the method how i done it actually,
i have tested the last hours the solution with sql lite on application
server

the Solution:

3 Mysql Server ( 1 more to handle the big load ) (1 Master, 2 Slaves)
mysql replication
10 Applikation Server ( get today two more from my hoster :-) )
on Application Serer running php and sql lite
by the first load from mysql server the big query get synchronised with
the lokal sql lite and write into the database
entrys are about 6 hour valid, after then the server get the new list.

performance looks nice:
total load time: between 0.03-0.09 Seconds

but i found another problem by the time i worked on the server
application server can create images and thumbs of them in various sizes
(gd lib etc.)
then this server open a ftp connection ( ftp_connect().... ) to a global
data & storage server
the data server has just running ftp so i must created the thumbs on
application server and move all files to the data server:
question:
*can php handle some compression with ftp* ? so that i can move some
more data?

chirs


Michael A. Peters schrieb:
> workerholic(a)studysite.eu wrote:
>> hi andrew i think you understand my problem a little,
>> but if 100 user load this query at the same time, the two mysql
>> server had a lot to do!
>> so i think to cache this query as xml to the application server local
>> make thinks faster,
>> but, i would like to have the same performance to read this xml
>> document as read the query from mysql server...
>> i dont know why php is so slow to read the xml file...
>
> Are you saving to file or caching as a query result?
> Also note that you can cache an array of rows (at least with APC but I
> suspect memcache as well) - say my_fetch(key) is you function to fetch
> from cache and my_store(key,data,life) is your function to store.
>
> $result = my_fetch('big_query');
> if (! $result) {
> $sql = 'your query';
> $rs = mysql_query($sql);
> while ($row = mysql_fetch_object($rs)) {
> $result[] = $row;
> }
> my_store('big_query',$result,21600);
> }
>
> No xml involved and you can loop through the results.
>
> If you'd rather do it as xml, you can cache the xml as a string and
> then fetch it, importing it into a DOM or whatever to extract your
> results.
>
> $xml = my_fetch('queryResultAsXML');
> if (! $xml) {
> generate xml and cache it
> }
> $dom = new DOMDocument('1.0','utf-8');
> $dom->loadXML($xml);
>
> Not sure what you are doing, apoligize if these suggestions are
> useless or already considered.
>

From: Phpster on 10 Jul 2009 18:52




On Jul 10, 2009, at 5:25 PM, "workerholic(a)studysite.eu" <workerholic(a)studysite.eu
> wrote:

> first thanks to all who have read ;-)
>
> your solution looks like the method how i done it actually,
> i have tested the last hours the solution with sql lite on
> application server
>
> the Solution:
>
> 3 Mysql Server ( 1 more to handle the big load ) (1 Master, 2
> Slaves) mysql replication
> 10 Applikation Server ( get today two more from my hoster :-) )
> on Application Serer running php and sql lite
> by the first load from mysql server the big query get synchronised
> with the lokal sql lite and write into the database
> entrys are about 6 hour valid, after then the server get the new list.
>
> performance looks nice:
> total load time: between 0.03-0.09 Seconds
>
> but i found another problem by the time i worked on the server
> application server can create images and thumbs of them in various
> sizes (gd lib etc.)
> then this server open a ftp connection ( ftp_connect().... ) to a
> global data & storage server
> the data server has just running ftp so i must created the thumbs on
> application server and move all files to the data server:
> question:
> *can php handle some compression with ftp* ? so that i can move some
> more data?
>
> chirs
>

You could zip the files together, but on the whole images don't
compress much. Setting the FTP process to run on a cron would be the
best way to avoid doing a lot of transfers at one time.


Bastien


First  |  Prev  | 
Pages: 1 2 3
Prev: How to stop E_DEPRECATED messages in the PHP log?
Next: How to Install Roadsend Compiler on Fedora