What's the ideal way to implement this PHP/XML function in a website?

Question

Hi,

I have this code written up to snatch some weather data and make it available on my website:

if( ! $xml = simplexml_load_file('http://www.weather.gov/data/current_obs/KBED.xml') ) 
{ 
    echo 'unable to load XML file'; 
} 
else 
{ 
  $temp = $xml->temp_f.' Degrees';
  $wind = $xml->wind_mph;
  $wind_dir = $xml->wind_dir;
  $gust = $xml->wind_gust_mph;
  $time = $xml->observation_time;
  $pres = $xml->pressure_in;
  $weath = $xml->weather;
}

And then I just echo them out inside the tags I want them inside. My site is low traffic, but I'm wondering what the "best" way is to do something like this if I were to spike way up in traffic. Should I write those variables I want into a database every hour (when the XML is refreshed) with a cron job to save pinging the server each time, or is that not bad practice? I understand this is a bit subjective, but I have no one else to ask for "best ways". Thanks!!

Answer 1

• Set up a cron job to periodically fetch the XML document, parse it and store the variables in a database.
• When a page is requested, fetch the variables from the database and render your page.
• It is a good idea to store the timestamp of the last update in the database as well, so that you can tell when the data is stale (because the weather website is down or so).

This setup looks very reasonable to me.

Answer 2

You could cache the output of the external site, and let it renew itself say every 5-10 seconds. That would kill the impact of a lot of 'pings' from your site. It really depends on how important timing accuracy is to your customer/client.

In a high traffic situation I would have a separate script that runs a a daemon or cron job and fetches the weather every specified interval, and overwrites the public website page when done. That way, you've not to worry about caching as it's done by a background task, your visitors are merely accessing a static page from the web server. That also avoids or at least minimises the need to incorporate a database into the equation, and is fairly light-weight.

On the downside, it does create a second point of failure and could be pretty useless if the information needs to be accurate to the time of page access.

Answer 3

I would suggest the following:

• When you first get the content of the xml, parse it, and serialise it to a file, with a timestamp attached to the file in some way (perhaps as part of the serialised data structure)

• Every time the page loads, grab that serialised data, and check the timestamp. If it's passed a certain point, go and grab the xml again and cache it, making sure to update the timestamp. If not, just use that data.

That should work, means you only have to go get the xml occasionally, and also, once the cache has expired, you don't have the waste of going and getting it regularly even though no-one is visiting (since it is only updated on a request).