Using Performance Timeline
The Performance Timeline standard defines extensions to the Performance
interface to support client-side latency measurements within applications. The standard also includes interfaces that allow an application to be notified when specific performance events occur. Together, these interfaces can be used to help identify an application's performance bottlenecks.
Performance extensions
Performance Timeline extends the Performance
object with three methods that provide different mechanisms to get a set of performance records (metrics)
, depending on the specified filter criteria. The following example show the usage of these methods getEntries()
, getEntriesByName()
and getEntriesByType()
.
function log(s) {
const o = document.getElementsByTagName("output")[0];
o.innerHTML += `${s} <br>`;
}
function doWork(n) {
for (let i = 0; i < n; i++) {
const m = Math.random(); // This is an example of work taking some time
}
}
function printPerfEntry(pe) {
log(`name: ${pe.name}`);
log(`entryType: ${pe.entryType}`);
log(`startTime: ${pe.startTime}`);
log(`duration: ${pe.duration}`);
}
function printPerformanceEntries() {
if (performance.mark === undefined) {
console.error("The property performance.mark is not supported.");
return;
}
// Create some performance entries via the mark() and measure() methods
performance.mark("Begin");
doWork(50000);
performance.mark("End");
doWork(50000);
performance.measure("Measure1", "Begin", "End");
// Use getEntries() to iterate all entries
performance.getEntries().forEach((entry, i) => {
log(`All Entry[${i}]`);
printPerfEntry(entry);
});
// Use getEntries(name, entryType) to get specific entries
performance
.getEntries({ name: "Measure1", entryType: "measure" })
.forEach((entry, i) => {
log(`Begin and Measure [${i}]`);
printPerfEntry(entry);
});
// Use getEntriesByType() to get all "mark" entries
performance.getEntriesByType("mark").forEach((entry, i) => {
log(`Mark only [${i}]`);
printPerfEntry(entry);
});
// Use getEntriesByName() to get all "mark" entries named "Begin"
performance.getEntriesByName("Begin", "mark").forEach((entry, i) => {
log(`Begin and Mark [${i}]`);
printPerfEntry(entry);
});
}
PerformanceEntry interface
The PerformanceEntry
interface encapsulates a single performance entry i.e. a single performance metric. This interface has four properties and a JSON serializer (toJSON()
. The following example shows the use of these properties.
function printPerformanceEntry(ev) {
const properties = ["name", "entryType", "startTime", "duration"];
// Create a few performance entries
performance.mark("Start");
doWork(50000);
performance.mark("Stop");
performance.measure("measure-1");
performance.getEntries().forEach((perfEntry, i) => {
log(`PerfEntry[${i}]`);
properties.forEach((prop) => {
// Check each property in window.performance
const supported = prop in perfEntry;
log(`… ${prop} = ${supported ? perfEntry[prop] : "Not supported"}`);
});
});
}
This interface also includes a toJSON()
method that returns the serialization of the PerformanceEntry
object. The following examples show the use of this method.
function perfEntryToJSON() {
// Create a few performance entries
performance.mark("mark-1");
performance.mark("mark-2");
performance.measure("meas-1", "mark-1", "mark-2");
const peList = performance.getEntries();
const pe = peList[0];
if (pe.toJSON === undefined) {
log("PerformanceEntry.toJSON() is NOT supported");
return;
}
// Print the PerformanceEntry object
const json = pe.toJSON();
const s = JSON.stringify(json);
log(`PerformanceEntry.toJSON = ${s}`);
}
Performance Observers
The performance observer interfaces allow an application to register an observer for specific performance event types, and when one of those event types is recorded, the application is notified of the event via the observer's callback function that was specified at the time, the observer was created. When the observer (callback) is invoked the callback's parameters include a performance observer entry list
that only contains observed performance entries
. That is, the list only contains entries for the event types that were specified when the observer's observe()
method was invoked.
The following example shows how to register two observers: the first one registers for several event types and the second observer only registers for one event type.
function PerformanceObservers() {
// Create observer for all performance event types
const observeAll = new PerformanceObserver((list, obs) => {
// Print all entries
list.getEntries().forEach((entry) => {
printPerfEntry(entry);
});
// Print entries named "Begin" with type "mark"
list.getEntriesByName("Begin", "mark").forEach((entry) => {
printPerfEntry(entry);
});
// Print entries with type "mark"
list.getEntriesByType("mark").forEach((entry) => {
printPerfEntry(entry);
});
});
// Subscribe to all performance event types
observeAll.observe({
entryTypes: [
"frame",
"mark",
"measure",
"navigation",
"resource",
"server",
],
});
// Create observer for just the "mark" event type
const observeMark = new PerformanceObserver((list, obs) => {
// Should only have 'mark' entries
list.getEntries().forEach((entry) => {
printPerfEntry(entry);
});
});
// Subscribe to only the 'mark' event
observeMark.observe({ entryTypes: ["mark"] });
}
function printPerfEntry(pe) {
log(`name: ${pe.name}`);
log(`entryType: ${pe.entryType}`);
log(`startTime: ${pe.startTime}`);
log(`duration: ${pe.duration}`);
}
The performance observer entry list
interface has the same three getEntries*()
methods as the Performance
interface and these methods are used to retrieve observed performance entries within the observer callback. These methods have been used in the above stated example.