Efficient handling of IO in PHP - php

Question

I'm using php, and handling file operation. My server need to response for every client request(concurrently minimum 5000 clients), My server open a xml file and convert xml to php array and will do some calculations and response as a json file, For this i'm using below code
$xmlstring = file_get_contents("../api/rate.xml");
$xml = simplexml_load_string($xmlstring);
$rate_json = json_encode($xml);
$rate_array = json_decode($rate_json, true);
$return_rates = array();
$return_rates['Baserates'] = $rate_array['Baserates'];
/* Here i will do some process and create $return_rates array */
$return_rates = json_encode($return_rates);
echo $return_rates;
This code producre resulat as i need, but i'm getting some times 500 internal server error, because of issue of handling IO(My server people saying this issue for internal server error). When many concurrent access happens in reading the file this issue happening, Please help any once to solve this issue.
XML File will be produce by third party application. For every second i'm receiving this XML file from one of my third party application.

Answer 1

You need to execute this code only when rate.xml changed. For first time, check if rate.xml content has changed by using hash value. If hash value is not same with last checked hash value then execute following code and store json file in local storage and new hash value.
$xmlstring = file_get_contents("../api/rate.xml");
$xml = simplexml_load_string($xmlstring);
$rate_json = json_encode($xml);
Otherwise you just need to read local json file and decode it. Or you can simply read from $xml without call to json_encode. Or you can request third party app developer to add JSON output instead of XML.
Also instead of file_get_contents(), you can use cUrl and sent HTTP HEAD first to find if rate.xml size changed instead of directly read rate.xml contents. When they do changed then you can call HTTP GET to retrieve their content.
The goal is to minimize network I/O and file I/O. Try cache rate.xml as long as possible in local storage or RAM (try Redis or Memcached)