System Highlights |
Client Connections:
|
This is an instantaneous snapshot of the client connection count.
|
Client Transactions Per Second:
|
This is an instantaneous snapshot of client ops/sec taken every five
minutes over a ten second window.
|
HTTP Document Hit Rate:
|
The blue curve shows the 5 minute average cache hit rate, including
fresh hits, and hits successfully revalidated with the origin server.
The orange curve shows the percentage of hits that were stale, but
revalidated successfully with the origin server.
|
RAM Cache Read I/O Hit Rate:
|
The blue curve shows the 5 minute percentage of all cache read I/Os that
were satisfied from the RAM cache.
|
Errors & Aborts (Percentage):
|
This illustrates the 5 minute average percentage of errors (blue) and
aborts (orange). Aborts are typically less than 5 percent in conditions
of good network connectivity.
|
HTTP Cache Hit Latency:
|
The 5 minute average latency (in milliseconds) for cache hits. Overall
cache hit latency is shown in blue. Fresh cache hits are shown in
orange. The hit latency is affected by cache load, object size, and
network conditions.
|
HTTP Cache Miss Latency:
|
The 5 minute average latency (in milliseconds) for cache misses. Overall
miss latency is shown in blue. Misses that were previously not cached,
but are now, are shown in orange.
|
CPU Available:
|
This illustrates the instantaneous percentage of all CPUs that are not
actively doing computation, broken into non-busy I/O wait (blue), and
non-busy, non-I/O idle time (orange).
|
<@record proxy.config.server_name> Memory Usage:
|
The amount of memory used by the <@record proxy.config.server_name> process, as
reported by the operating system. Total virtual memory space is shown
in blue, and the current physically resident size is shown in orange.
|
Transaction Breakdown |
Hits & Misses (Percentage):
|
This illustrates the 5 minute average percentage of hits (blue)
and misses (orange).
|
Errors & Aborts (Percentage):
|
This illustrates the 5 minute average percentage of errors (blue) and
aborts (orange). Aborts are typically less than 5 percent in conditions
of good network connectivity.
|
MSIE Requests (Percentage):
|
This illustrates the 5 minute average percentage of requests coming from
Microsoft Internet Explorer browsers (blue), and the percentage of requests
coming from MSIE browsers that have an IMS or no-cache header (orange).
|
Bandwidth Savings:
|
This shows the fraction of bytes served from the cache as opposed to
from the network. This is sometimes referred to as a "byte hit rate"
or "hit rate by bytes". The five minute average is shown in blue, and
samples of 10 sec averages reported by the manager are shown in orange.
|
HTTP Hits & Misses (Count):
|
This illustrates the total number of hits (blue) and misses (orange) over
the last 5 minute interval.
|
HTTP Errors & Aborts (Count):
|
This illustrates the total number of errors (blue) and aborts (orange)
over the last 5 minute interval.
|
Transaction Latency |
HTTP Cache Hit Latency:
|
The 5 minute average latency (in milliseconds) for cache hits. Overall
cache hit latency is shown in blue. Fresh cache hits are shown in
orange. The hit latency is affected by cache load, object size, and
network conditions.
|
HTTP Cache Miss Latency:
|
The 5 minute average latency (in milliseconds) for cache misses. Overall
miss latency is shown in blue. Misses that were previously not cached,
but are now, are shown in orange.
|
HTTP Error Latency:
|
The 5 minute average latency (in milliseconds) for errors originating
at the proxy. These are typically DNS failures and connect timeouts.
These do not count server-originating errors, which count as misses.
|
HTTP Abort Latency:
|
The 5 minute average latency (in milliseconds) for requests aborted
by the client.
|
Object Store Metrics |
Cache Usage:
|
The size of the cache in megabytes currently in use is shown in blue. The orange line
depicts the total size in megabytes available for caching.
|
RAM Cache Usage:
|
The blue curve shows the current active size of the RAM cache
in megabytes. The orange curve depicts the size of the RAM cache in megabytes
that are currently locked.
|
DNS Cache Usage:
|
The blue curve shows the total number of DNS requests over the last
5 minutes.
|
CPU Utilization |
CPU Available:
|
This illustrates the instantaneous percentage of all CPUs that are not
actively doing computation, broken into non-busy I/O wait (blue), and
non-busy, non-I/O idle time (orange).
|
CPU Busy:
|
This illustrates the instantaneous percentage of all CPUs that are
actively processing, either in the kernel (blue), or in
applications (orange).
|
Connection Counts |
Client Connections:
|
This is an instantaneous snapshot of the client connection count.
|
Origin Server Connections:
|
This is an instantaneous snapshot of the origin server connection count.
|
Parent Proxy Connections:
|
This is an instantaneous snapshot of the parent proxy connection count.
|
Network Statistics |
TCP Throughput:
|
This is the total TCP network I/O, in bits/second. The data is
measured by netstat.
|
TCP Segment Transmitted:
|
The blue curve graphs the number of TCP segment transmitted. The orange curve
demarks the portion of the segment which were re-transmitted due to
packet loss or timeout. Under good network conditions, there should
be very few retransmissions.
|
TCP Connect Rate:
|
This shows the 5 minute average TCP connect rate. The blue curve shows
the number of incoming connects/sec, while the orange curve shows the
number of outgoing connects/sec.
|
TCP ESTABLISHED Connections:
|
The instantaneous count of the TCP connections in the ESTABLISHED
state. An ESTABLISHED connection is normally exchanging data.
|
TCP CLOSE_WAIT Connections:
|
The instantaneous count of the TCP connections in the CLOSE_WAIT state.
CLOSE_WAIT connections have been closed by a remote peer, but the
<@record proxy.config.server_name> has not closed yet.
|
TCP FIN_WAIT_1 Connections:
|
The instantaneous count of the TCP connections in the FIN_WAIT_1 state.
FIN_WAIT_1 connections have been closed by the <@record proxy.config.server_name>, but no
acknowledgement of the close has yet been received from the remote peer.
|
TCP FIN_WAIT_2 Connections:
|
The instantaneous count of the TCP connections in the FIN_WAIT_2 state.
FIN_WAIT_2 connections have been closed by the <@record proxy.config.server_name>, and
acknowledged by the remote peer, but the other party has not yet closed.
|
TCP TIME_WAIT Connections:
|
The instantaneous count of the TCP connections in the TIME_WAIT state.
TIME_WAIT connections are fully closed (<@record proxy.config.server_name> initiated the
close), but will still be consuming OS table space for 2MSL seconds.
|
TCP LAST_ACK Connections:
|
The instantaneous count of the TCP connections in the LAST_ACK state.
LAST_ACK connections have been closed first by the remote peer, then
closed by the <@record proxy.config.server_name>. The OS
is awaiting the acknowledgement of the final close.
|
<@record proxy.config.product_name> Memory Usage |
<@record proxy.config.server_name> Memory Usage:
|
The amount of memory used by the <@record proxy.config.server_name> process, as
reported by the operating system. Total virtual memory space is shown
in blue, and the current physically resident size is shown in orange.
|
<@record proxy.config.manager_name> Memory Usage:
|
The amount of virtual memory used by the <@record proxy.config.manager_name> process, as
reported by the operating system.
|
Real Proxy Memory Usage:
|
The amount of virtual memory used by the Real Audio proxy process, as
reported by the operating system.
|
Media Protocols |
RNI Client/Server Connections:
|
This is an instantaneous snapshot of the RNI client/server connection count.
The blue curve and the orange line represent the client/server connection count, respectively.
|
RNI Hit Rate:
|
The blue curve shows the 5 minute average RNI hit rate.
|
RNI Bandwidth Savings:
|
This shows the RNI bandwidth savings percentage.
|
WMT Client/Server Connections:
|
This is an instantaneous snapshot of the WMT client/server connection count.
The blue curve and the orange line represent the client/server connection count, respectively.
|
WMT Unique Live Stream Connections:
|
This is an instantaneous snapshot of the WMT current unique live stream and on-demand connection count.
The blue curve and the orange line represent the live stream/on-demand connection count, respectively.
|
WMT Hit Rate:
|
The blue curve shows the 5 minute average WMT hit rate.
|
WMT Bandwidth Savings:
|
This shows the WMT bandwidth savings percentage.
|
QT Client/Server Connections:
|
This is an instantaneous snapshot of the QT client/server connection count.
The blue curve and the orange line represent the client/server connection count, respectively.
|
QT Unique Live Stream Connections:
|
This is an instantaneous snapshot of the QT current unique live stream connection count.
|
QT Hit Rate:
|
The blue curve shows the 5 minute average QT hit rate.
|
QT Bandwidth Savings:
|
This shows the QT bandwidth savings percentage.
|
MPEG4 Client/Server Connections:
|
This is an instantaneous snapshot of the MPEG4 client/server connection count.
The blue curve and the orange line represent the client/server connection count, respectively.
|
MPEG4 Unique Live Stream Connections:
|
This is an instantaneous snapshot of the MPEG4 current unique live stream connection count.
|
MPEG4 Hit Rate:
|
The blue curve shows the 5 minute average MPEG4 hit rate.
|
MPEG4 Bandwidth Savings:
|
This shows the MPEG4 bandwidth savings percentage.
|
Cache Disk Statistics |
Number of Cache Read:
|
The graph shows the number of cache reads per second.
|
Number of Cache Writes:
|
The graph shows the number of cache writes per second.
|
Kilobytes of Cache Read:
|
The graph shows the number of kilobytes read from the cache per second.
|
Kilobytes of Cache Writes:
|
The graph shows the number of kilobytes written to cache per second.
|
Internal Statistics |
Network Reads:
|
This shows the total number of network reads performed (blue) over the
last 5 minutes, along with the number of network reads that were wasted,
returning no data (orange).
|
Network Writes:
|
This shows the total number of network writes performed (blue) over the
last 5 minutes, along with the number of network writes that wrote no
data (orange).
|
Duty Cycles |
Server Duty Cycle:
|
Duty cycle for <@record proxy.config.product_name> components in seconds.
|
MRTG Runtime:
|
Wall-clock time elapsed while gathering statistics and generating graphs.
The blue curve depicts the wall-clock time elapsed while completing the
previous MRTG cycle, and should be well below the sampling
interval (5 minutes).
The orange curve represents the wall-clock time elapsed while completing
just the graph generation phase of the previous MRTG cycle.
NOTE: arbitrary delays are added to prevent MRTG consuming 100% of CPU time.
|
|
Based on:
Tobias Oetiker
<oetiker@ee.ethz.ch>
and Dave Rand
<dlr@bungi.com>
|