It can take a while depending on the size of the document..please wait
Discuto
0 days left (ends 29 Aug)
description
Further info
LATEST ACTIVITY
LEVEL OF AGREEMENT
MOST DISCUSSED PARAGRAPHS
LATEST COMMENTS
MOST ACTIVE USERS
4.1.4Detection of an HTTP proxy
P104
An HTTP proxy is a middlebox that is inserted into the path for end users' HTTP connections, which may be used to filter or modify traffic. A HTTP proxy can be transparent or otherwise hidden.
Add comment
P105
A transparent proxy is a middlebox deployed by the IAS provider which acts as an intermediary between the client and the target web server. Typically, the ISP routes HTTP traffic via the proxy without user action or knowledge.
Add comment
P106
A transparent HTTP proxy might be detected by checking the HTTP headers for proxy specific content (HTTP_VIA, VIA, FORWARDED, CLIENT-IP…).
Add comment
P107
The HTTP (TRACE) request headers could also be checked for modification between the client and the server and if the intercepting proxy does a DNS lookup on a fake host header.[3] A hidden proxy could be detected by a cache test[4].
Add comment
P108
Some HTTP proxies can be detected by connecting to a target domain and checking that the web resource is available and verifying that the content is identical to the content received over a proxy outside the ISPs control[5].
Add comment
P109
Finally, it may be possible to detect a HTTP proxy by inspecting properties of the sent traffic (e.g. TTL-flag of the IP packet).
Add comment
P110
The purpose of these measurements is both to suggest other indicators of performance closer to the user experience, which could give to the consumer some easily understandable criteria to help him to take enlightened decisions, and to detect the prioritisation and/or throttling of specific applications. These traffic management practices may be detected by measuring some of the KPIs described below and comparing the results based on the following variations:
Add comment
P111
• Comparison of the same KPIs related to similar applications for the same IAS subscription,
Add comment
P112
• Comparison of the KPIs for the same application using an equivalent subscription from another IAS provider, and/or
Add comment
P113
• Comparison of the KPIs for the same application and the same IAS subscription but using a VPN (see section 6.4).
Add comment
P114
These measurements can be performed on a regular basis for selected applications, websites or platforms or in targeted situations as needed.
Add comment
4.2.1Web browsing
P115
In order to assess web browsing quality of service, the time (in seconds) needed to load a web page for the first time could be a good indicator. A first loading experience can be simulated by the measurement tools e.g. by forcing local cache clearance or to configure the web server appropriately.
Add comment
P116
One option can be to use a normalised reference page (e.g. ETSI (mobile) Kepler page)[6] from a dedicated web server. This avoids bias by fixing the page size and associated page elements, and removes the dependency on the performance of a web server on the open internet. A second approach consists of measuring the time taken to reach the HTML and referenced resources from a page of a real website. The panel of chosen websites can include popular websites, government website, etc.
Add comment
P117
A significant increase of the time needed to load the web page can be an indication of throttling or prioritisation.
Add comment
4.2.2Video streaming
P118
In order to assess video streaming QoS, several options could be considered. The first approach is to simulate a data stream comparable to a normal video streaming session on which standard measurements of IAS quality can be carried out (bitrate, latency, etc.).
Add comment
P119
The second approach consists of launching a video streaming session on an existing public streaming platform and measuring some important characteristics:
Did you know you can vote on comments? You can also reply directly to people's comments.