|
The following table summarizes the meaning of the terms used in the
summary reports:
| Term |
Meaning |
| Hit |
Any request for a document, a script or an
image, which causes the server to respond in some way. If a HTML page
is requested this may cause further requests for images or other
embedded objects to be sent to the server. |
| Files |
Amount of data sent on behalf of requests
for documents, images, applets, and all other type of HTML objects (sum
of all Code 200 OK responses). |
| 304 Responses |
Sum of all Code 304 NotModified
responses. This response is generated if a document's content hasn't
been changed since the last time this document was transferred to a
visitor. |
| Sites |
Number of unique hosts. Note: While sites
denotes the unique sites per day in the short summary, it reflects the
number of unique sites per month in the full summary. This may cause
different values to show up in the short and in the full summary. |
| Unique URLs |
Number of unique URLs. Beside of HTML pages,
images, CGI scripts, applets, and all other HTML objects are counted
also as unique URLs. |
| KB requested |
Total amount of data requested. This is the
sum of KB transferred and KB saved by cache
(see below). |
| KB transferred |
Total amount of data really sent by the
server (the sum of the sizes of all Code 200 OK
responses). |
| KB saved by cache |
Total amount of data saved by cache because
of proxy servers or cache mechanisms in the user agent. The sum of this
value and the data transferred is the number shown in the KB
requested field. Because http-analyze can
determine the size of a page only if this page has been already
transferred at least once in the same summary period, the numbers shown
in the KB saved by cache and KB requested
fields are only approximations of the real values. |
| Country |
Country a host resides in if it could be
determined by the two-letter Internet
Country Code. Normally, this country code is used as
the top-level domain in a hostname, except in the US, where three- and
four-letter top-lebel domains are used to reflect the type of
organization (such as .com for commercial or .edu
for educational institiutions). If the web server couldn't resolve the
numerical IP address into a hostname, http-analyze
uses this IP address to account for the total number of unique hosts,
but later comprises such hosts under the term Unresolved.
Usually, the share of unresolved IP addresses on all hostnames is about
15 to 30%. Currently, http-analyze does not try to
resolve numerical IP addresses. |
IMPORTANT!
When reviewing http-analyze, logfiles, please repeat the following
prayer in the usual attitude of your locale at least five times:
The statistics generated by http-analyze DOES allow me to
determine the access load of my web server. The reports DOES NOT allow
me to figure out
-
the number of visitors of a page or the server,
-
page views (whatever this is)
-
the user's behaviour,
-
the time a user looks on my page,
-
the sudden death of the user while looking at my homepage.
The reasons why http-analyze does not produce exact values for
the total number of visitors or figures about their behaviour, are
caching mechanisms and the nature of the HTTP protocol:
-
A proxy server (and also a navigator) may cache the data requested from
an outside web server. If a user requests a page, which is already in
the proxy's (or navigator's) cache, it will be served from the local
cache rather than being transmitted from the web server again if the
content of the page has not changed since the last transmission. The
proxy (navigator) asks the web server, whether the document has changed
in the meantime. If it hasn't, it causes the server to send a Code
304 NotModified response). While this helps to save bandwidth,
it prevents an exact statement about the number of visitors of a
server, since mostly all online services and many other big
installations use proxy servers with big caches and so thousands of
users can visit your page causing a few hits only.
-
The HTTP protocol is stateless. You can't figure out when a visitor
jumps onto your page and when she/he leaves it and why.
http-analyze has been developed mainly to be able to classify
virtual web servers depending on their traffic and usage profile for
accounting reasons. The most important weight for this classification
is given the amount of data sent (KB transferred), which
reflects exactly, what has been put on the wire on behalf of requests.
The remaining values let you determine the access load of your web
server in some way. All other interpretations of the resulting numbers
are at least questionable.
|
