I was part of a discussion about this this topic. The thread participants decided that GlideAjax should be favored over client-side GlideRecord, based on recommendations on the SN wiki. I decided to see for myself.
I created a GlideAjax script to fetch a manager:
function onLoad() {
var ga = new GlideAjax("GMF_BenchmarkSample");
ga.addParam("sysparm_name", 'getManagerId');
ga.addParam("sysparm_userId", g_form.getValue('caller_id'));
ga.getXML(callBack);
}
function callBack(response){
var answer = response.responseXML.documentElement.getAttribute("answer");
alert(answer);
}
On the back end I used this script include:
var GMF_BenchmarkSample = Class.create();
GMF_BenchmarkSample.prototype = Object.extendsObject(AbstractAjaxProcessor, {
//Log the error using server side error logging
getManagerId: function () {
var userId = this.getParameter('sysparm_userId');
var gr = new GlideRecord('sys_user');
if(gr.get(userId)){ return gr.manager;}
return null;
},
});
Then I created a GlideRecord version to do the same thing:
function onLoad() {
var rec = new GlideRecord('sys_user');
rec.addQuery('sys_id', g_form.getValue('caller_id'));
rec.query(recResponse);
}
function recResponse(rec) {
while (rec.next()) {
alert(rec.manager);
}
}
Based on the advice on the wiki, the GlideRecord version should perform worse ostensibly due to having to load all of the properties on the user record. However I found this:
Since the client GlideRecord is considerably easier to work with and only causes a negligible performance hit, I'm switching to using it in the future.
(The test was executed multiple times, each time against the same INC record, with a margin of error of about 6ms. Occasionally the GlideRecord script even outperformed the GlideAjax script. The outliers happened when the wait time became 4-5 time longer than normal at random intervals.)
