Network Working Group M. McCall
Internet-Draft Akamai
Intended status: Standards Track June 16, 2012
Expires: December 18, 2012
HTTP Extension to provide timing data for performance measurement
draft-mccall-httpbis-timing-measurements-00
Abstract
This memo presents a method for allowing an HTTP server to gather
relevant performace data from compliant user-agents. Though the HTML
5 standard supplies Nav-Timings there is no standard way for that
data to be communicated back to an origin. The intention of this
document is to describe a method in which that can be done cleanly
and scalably.
Status of this Memo
This Internet-Draft is submitted in full conformance with the
provisions of BCP 78 and BCP 79.
Internet-Drafts are working documents of the Internet Engineering
Task Force (IETF). Note that other groups may also distribute
working documents as Internet-Drafts. The list of current Internet-
Drafts is at http://datatracker.ietf.org/drafts/current/.
Internet-Drafts are draft documents valid for a maximum of six months
and may be updated, replaced, or obsoleted by other documents at any
time. It is inappropriate to use Internet-Drafts as reference
material or to cite them other than as "work in progress."
This Internet-Draft will expire on December 18, 2012.
Copyright Notice
Copyright (c) 2012 IETF Trust and the persons identified as the
document authors. All rights reserved.
This document is subject to BCP 78 and the IETF Trust's Legal
Provisions Relating to IETF Documents
(http://trustee.ietf.org/license-info) in effect on the date of
publication of this document. Please review these documents
carefully, as they describe your rights and restrictions with respect
to this document. Code Components extracted from this document must
include Simplified BSD License text as described in Section 4.e of
the Trust Legal Provisions and are provided without warranty as
McCall Expires December 18, 2012 [Page 1]
Internet-Draft Timing Measuements June 2012
described in the Simplified BSD License.
Table of Contents
1. Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2. Supported Client Measurements . . . . . . . . . . . . . . . . 5
3. Delivering Measurement Data . . . . . . . . . . . . . . . . . 6
4. Illustrative Use Case . . . . . . . . . . . . . . . . . . . . 7
5. Proposed HTTP protocol extension . . . . . . . . . . . . . . . 9
6. Security Considerations . . . . . . . . . . . . . . . . . . . 10
7. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 11
8. Closing Thoughts . . . . . . . . . . . . . . . . . . . . . . . 12
9. References . . . . . . . . . . . . . . . . . . . . . . . . . . 13
9.1. Normative References . . . . . . . . . . . . . . . . . . . 13
9.2. Informative References . . . . . . . . . . . . . . . . . . 13
Author's Address . . . . . . . . . . . . . . . . . . . . . . . . . 14
McCall Expires December 18, 2012 [Page 2]
Internet-Draft Timing Measuements June 2012
1. Overview
One of the primary goals of the HTTP 2.0 is to improve performance of
web pages. While there have been many performance improvements in
the entire web stack since the first iteration of HTTP - from server,
to network, to browser - there are now important efforts being made
to improve the protocol which delivers much of the world's content
every day. Indeed, it's no surprise that having a faster website has
shown to improve customer conversion rates and user engagement, and
with richer websites becoming more prevalent, improving performance
is a natural next step in the evolution of the protocol. While there
are many different proposals on how best to speed up HTTP, this
document focuses on quantifying the performance of an HTTP session.
In its current state, HTTP has no means to expose performance metrics
for an HTTP session. Because of this, today, most web page
performance measurement is done one of two ways:
Synthetic transactions
Real-user Measurement
Synthetic transactions tend to rely on a instrumented browser that
has the ability measure DNS resolution time, first byte time, object
download time, DOM processing time, and other browser events. There
are many synthetic transaction options currently available, ranging
from 3rd-party solutions provided by companies like Keynote [Keynote]
and Gomez [Gomez] , to open-source projects like WebPageTest [WPT] .
While these solutions are valuable, they do have limitations. In
particular, most these measurements are performed in controlled,
"clean room" environment. These "clean rooms" often have better-
performing networks, browsers, and computers than a real user might
have, and thus, tend to provide a best-case measurement of a site's
performance.
Real-user measurement is normally performed by instrumenting the
target webpage with JavaScript to capture measurements similar to
those available in synthetic transactions. Recently, the W3C has
submitted a few specification drafts [Nav-Timing] which provide an
interface for browsers to expose performance timing data, and most
modern browsers support the Navigation Timing interface. While the
measurements available using browser-side RUM are extremely valuable
in quantifying an end-user experience, this solution also comes with
limitations. In particular, running JavaScript on the client does
not come without some performance impact. In addition, instrumenting
the webpage with timing measurements can potentially be non-trivial
and add to development time for web page designers.
McCall Expires December 18, 2012 [Page 3]
Internet-Draft Timing Measuements June 2012
This document outlines an alternative to both synthetic measurements
and real-user measurements in their current forms by adding new
request and response headers to the HTTP protocol. By coupling
client-side metrics into the HTTP session, end-user web page
performance can be easily quantified without the need and overhead of
JavaScript running in the client's browser, and without having to
rely on third parties to provide performance data. Additionally,
since the server is privy to the performance of the session, it is
possible for it to modify the content delivered accordingly.
At a high level, the HTTP Timing Measures Extension would work in a
manner similar to the following:
A new client request header, X-Accept-Measurement, is sent at the
initiation of an HTTP request.
A negotiation takes place between the server and the client to
determine which measurements will be sent by the client and a
timing interval for the measurements. The server requests
measurements by sending an X-Send-Measurement response header.
Assuming the negotiation is successful, the client sends
measurements after the timing interval has expired to the server
in an HTTP POST request with an X-Timing-Measures header.
We believe that adding this functionality to HTTP is important for a
variety of reasons. CDNs, in particular, compete through performance
attributes. However, there are a number of issues that make showing
value through performance data difficult: Web site owners tend to be
hesitant to add new JavaScript to their sites out of fear of slowing
the site down, and synthetic measurements don't truly show what real
users are experiencing. Even if a CDN were to transparently insert
measurement code into a web page, there are risks involved. By
adding measurements directly into HTTP, the playing field is leveled
by having the protocol itself provide the de-facto method of
measuring web site performance.
McCall Expires December 18, 2012 [Page 4]
Internet-Draft Timing Measuements June 2012
2. Supported Client Measurements
The W3C spec for Navigation Timing provides an excellent interface
for collecting base page metrics. However, it is limited to network
and browser events, and in those cases only for the base page as it
is served. AJAX-initiated events and object timings are not exposed.
The draft W3C spec for Resource Timing goes further, but to date has
no browser support.
In addition to the base measurements outlined in Navigation Timing,
the client should optionally support providing measurements for all
resources required to render the page within the timing interval
specified by the server. Having this level of measurement
granularity would allow one to collect measurements for objects that
are delivered outside of the base page measurement, and be useful in
illustrating end user experience.
At a minimum, clients should support the measurements outlined in the
Navigation Timing spec. Supporting other resource timings should be
optional, since it could conceivably generate a large number of
performance metrics to be sent back to the server.
McCall Expires December 18, 2012 [Page 5]
Internet-Draft Timing Measuements June 2012
3. Delivering Measurement Data
Once the client and server have negotiated the collection of the
measurements, data should be sent back to the collecting server via
an empty HTTP POST request, and encapsulated in an X-Timing-Measures
header.
However, there may be a case in which the amount of data to be sent
back to the server is too much to encapsulate in an HTTP header. An
HTTP POST also may not be desirable in cases when POSTs interfere
with application logic. In this case, an optional HTTP header,
X-Timing-Measures-Collector can be specified in the initial response
to indicate to the client where to send the data. If the X-Timing-
Measures-Collector header exists, data will be sent in the body of
the POST request, in the format previously described.
In some cases, it may be necessary to specify a Timing Measures
Collector on a per-page basis, in which case an optional meta element
can be added to an HTML page. Such a meta element might look like:
<meta http-equiv="Timing-Measures-Collector"
content="http://www.example.com/beacon.htm">
By default, the X-Timing-Measures-Collector attribute is null, and
data is POSTed back to the page as described earlier.
McCall Expires December 18, 2012 [Page 6]
Internet-Draft Timing Measuements June 2012
4. Illustrative Use Case
This use case demonstrates a successful performance measurement
negotiation and data posting. In it, we show how a user agent and
site (www.example.com) would negotiate and transfer measurement data.
Request1:
UA --> Server (www.example.com)
GET /index.html HTTP/2.0
Host: www.example.com
X-Accept-Measurement: allowed
Response1:
Server (www.example.com) --> UA
HTTP/2.0 200 OK
X-Send-Measurement:
dns;firstByte;domReady;domLoaded;onLoad;resources:TI:300000
We see that the first UA request has indicated that it is willing to
provide HTTP performance measurements (X-Accept-Measurement:
allowed), and that the server has replied with the measurements it is
requesting (X-Send-Measurement:
dns;firstByte;domReady;domLoaded;onLoad;resources:TI:300000). The
X-Send-Measurement response header sent by the server is followed
with a semicolon-delimited list of desired measurements, followed by
'TI', which is the timing interval of the desired measurement data
expressed as milliseconds.
Request2:
UA --> Server (www.example.com)
POST /index.html HTTP/2.0
Host: www.example.com
X-Timing-Measures:
dns:23;firstByte:120;domReady:1200;
domLoaded:1300;onLoad:1331;resources:1400
Response2:
Server (www.example.com) --> UA
HTTP/2.0 200 OK
We now see the UA sending back data to the initial server via an
empty HTTP POST request, including the X-Timing-Measures header
containing measurements that happened in the timing interval
requested by the server in the initial handshake. The header
contains key-value pairs of the requested measurements, where the
McCall Expires December 18, 2012 [Page 7]
Internet-Draft Timing Measuements June 2012
values are represented as milliseconds of the measurement's duration.
Depending on which measurements are supported by the browser, all or
some of the measurements requested by the server are returned.
McCall Expires December 18, 2012 [Page 8]
Internet-Draft Timing Measuements June 2012
5. Proposed HTTP protocol extension
To implement Timing Measures, additions of new HTTP headers are
required. The details of the request and response headers are
included as an initial starting point for discussions, rather than as
a final proposal.
Pseudo-BNF grammar for these headers might look like:
X-Accept-Measurement = "Accept-Measurement:" measure-resp
measure-resp = "accept|not-accepted"
X-Send-Measurement = "Send-Measurement:" measure-type:TI
measure-type = measurement
TI = timing-interval
X-Timing-Measures = "Timing-Measures:" measure-type:value
measure-type = measurement
value = value
X-Timing-Measures-Collector = "Timing-Measures-Collector:" collector-url
collector-url = collector
The Accept-Measurement, Send-Measurement headers are optional. If a
negotiation is successful, the Timing-Measures header is required.
The Timing-Measures-Collector header is optional. All measurement
values, as well as TI, are 32-bit integers expressed as milliseconds.
The collector-url is a string representing the URI data should be
sent back to.
Measurements are submitted via an HTTP POST request and must contain
the Timing-Measures header, and should not contain any data in the
body of the request. Measurements are sent as key-value pairs in the
form of measurement:value delimited by semicolons. If the Timing-
Measures-Collector header is sent or the relevant HTML meta tag
exists, the client must send its data to the specified collector-url
in the body of a POST request, using the data format specified above.
The user agent must discard the measurements after the Timing
Interval for the requested metrics expires.
McCall Expires December 18, 2012 [Page 9]
Internet-Draft Timing Measuements June 2012
6. Security Considerations
User-agent MUST follow 'same origin' considerations when sending
timing data
McCall Expires December 18, 2012 [Page 10]
Internet-Draft Timing Measuements June 2012
7. IANA Considerations
The new header fields
Accept-Measurement
Send-Measurement
Timing-Measures
Timing-Measures-Collector
should require IANA assignment
McCall Expires December 18, 2012 [Page 11]
Internet-Draft Timing Measuements June 2012
8. Closing Thoughts
Outside of adoption, the goal of this document is to get the
community thinking about measuring HTTP performance. While many of
the current HTTP 2.0 proposals have been designed to improve
performance, to date, none discuss quantifying the intended
improvements.
It is well known by the author that there are shortcomings and
unanswered questions in this document. In particular, there is a
lingering question about whether or not the timing measurements
should exist within the context of a server connection, or whether
they are best viewed within the scope of a user session. Another
question arises as to what should happen if the user agent closes the
connection before the timing interval has expired. With the optional
X-Timing-Measures-Collector header that can be used to send the
measurement results to another host, it is the belief of the author
that measurements should exist as part of a user session, and not
necessarily tied to a particular connection. Community consensus on
this matter is encouraged.
Another concern is that some of the measurements outlined in the
"Supported Client Measurements" section are more browser-specific
than some may be comfortable with. However, while network timing is
most obviously relevant for HTTP, it is also possible to infer
browser and hardware performance by looking at the time elapsed
processing and rendering the document. If a server had a full
performance profile of the user agent it was communicating with, it
could potentially serve different content or make other decisions
based on the UA's performance characteristics.
One glaring omission is the security of the proposed extension.
While performance data would seem to be of low value to malicious
users, at the very least the security of the extension should be
considered.
Finally, the grammar for the new headers needs to be properly
defined. The grammar outlined in the "Proposed HTTP Protocol
Extension" section is woefully incomplete and syntactically
incorrect, and is included only for illustrative purposes.
In closing, the intention is that these, as well as other
shortcomings in this document, are fleshed out by the community. The
more people thinking about these issues, the faster and better the
web will be.
McCall Expires December 18, 2012 [Page 12]
Internet-Draft Timing Measuements June 2012
9. References
9.1. Normative References
[RFC2119] Bradner, S., "Key words for use in RFCs to Indicate
Requirement Levels", BCP 14, RFC 2119, March 1997.
9.2. Informative References
[Keynote] "Keynote Systems", <http://www.keynote.com>.
[Gomez] "Gomez Networks", <http://www.gomeznetworks.com>.
[WPT] "WebPageTest", <http://www.webpagetest.org>.
[Nav-Timing]
"Nav-Timing Spec",
<http://www.w3.org/TR/navigation-timing/>.
McCall Expires December 18, 2012 [Page 13]
Internet-Draft Timing Measuements June 2012
Author's Address
Mike McCall
Akamai Technologies
McCall Expires December 18, 2012 [Page 14]