Measurement Variables
If you are using a PacketShaper ISP, see PacketShaper ISP
Measurement Variables.
Measurement variables specific to the compression feature
are in a separate table.
See Compression Measurement Variables.
Note: When you upgrade to a new PacketWise version,
you need to reset the
measurement data in order to access any new variables introduced in
this version.
PacketWise provides several sets of measurement variables.
The Common variables are available for all measurement types. Each measurement
type — link, partition, and class — also has its own set of
variables. The values for these variables are collected over the selected
interval. The following tables list the variables by type: Common,
Link, Partition, and Class.
Table of Common Variables
PacketWise stores link, partition, and class measurement data for the
variables in this table.
| Common Variables |
Description |
| avg-bps |
Usage, averaged over the interval, in bits per second |
| avg-pps |
Packets-per-second average rate
Formula: pkts / sample-interval-secs |
| bytes |
Number of bytes that passed through the unit during the specified
interval |
| element |
ID assigned to the class |
| guar-rate-allocs |
Count of allocation events that were allowed their guaranteed rates
during the measurement interval |
| guar-rate-fails |
Count of "denied" admission-control events when there wasn't enough
bandwidth to satisfy guaranteed rate demand. This is an indicator
of guaranteed rate failures and therefore can be used as a measurement
of user satisfaction. |
| kbytes |
Number of bytes transferred during the measurement interval, divided
by 1024 and rounded to the nearest kbyte |
| peak-bps |
Peak rate recorded for the link, partition, or class during the
specified interval; PacketWise determines the peak-bps by looking
at the rate recorded for the busiest one-second subinterval (that
is, the subinterval that had the highest rate) |
| peak-guar-rate-flows |
Guaranteed rate flow peak count |
| peak-tcp-conns |
Highest number of simultaneous TCP connections recorded for the
interval
Note: The peak number of TCP connections is not the same as the peak number of active flows. Some connections may not become active flows due to class policies, such as a class license or never-admit policy. |
| pkts |
Packet count — number of packets that passed through the unit
during the specified interval |
| sample-interval-msecs |
Time between measurement samples in milliseconds |
| sample-interval-overruns |
Interval overrun count — that is, the number of times the
measurement engine was too busy to be able to write the data to disk |
| sample-interval-secs |
Time between measurement samples (interval duration) in seconds |
| tcp-conn-aborts |
Number of TCP connections exited as aborted connections; for example,
HTTP stop button hits. This number gets incremented when a connection
is in progress and an RST is sent. (RSTs are sometimes used as a lazy
way to close a connection as an alternative to sending the full orderly
shutdown FIN, FIN-ACK, ACK sequence to close a TCP connection.) |
| tcp-conn-aborts% |
Percentage of TCP connections exited as aborted connections
Formula: tcp-conn-aborts / tcp-conn-exits |
| tcp-conn-exits |
Number of TCP connections exited. This number gets incremented
when PacketWise sees the end of a flow with an orderly shutdown (FIN,
FIN-ACK, ACK) or terminated by an RST. |
| tcp-conn-inits |
Number of TCP connections started. This number gets incremented
when PacketWise sees a SYN packet initiating a new flow. |
| tcp-conn-self-denies |
Number of PacketWise-initiated TCP connections exited because admission
control (refuse) was applied |
| tcp-conn-self-denies% |
Percentage of PacketWise-initiated TCP connection attempts exited
as denied
Formula: tcp-conn-self-denies / tcp-conn-exits |
| tcp-conn-server-ignores |
Number of TCP connections exited as ignored connections —
that is, the server never responded. This number gets incremented
when a connection is quarantined (which means the flow limit specified
in the flowlimit
policy was exceeded) or if either the SYN ACK or SYN ACK ACK are
not seen within a one-minute time-out period. |
| tcp-conn-server-ignores% |
Percentage of TCP connections exited as ignored connections
Formula: tcp-conn-server-ignores / tcp-conn-exits |
| tcp-conn-server-refuses |
Number of TCP connections refused by the server. This number gets
incremented when a SYN is refused using an RST. Generally this happens
when a server wants to deny a connection because it's too busy to
accept a new connection. |
| tcp-conn-server-refuses% |
Percentage of refused connections
Formula: tcp-conn-server-refuses / tcp-conn-exits |
| tcp-data-pkts |
Count of TCP data packets, including retransmits |
| tcp-early-retx-toss-pkts |
Count of TCP tossed retransmissions; PacketWise controls retransmission
time-outs for outbound retransmissions by discarding premature RTO
segments. |
| tcp-early-retx-toss-pkts% |
TCP toss rate
Formula: tcp-early-retx-toss-pkts / tcp-early-retx-toss-pkts + tcp-retx-pkts |
| tcp-efficiency% |
Percentage of bytes that were not retransmitted
Formula: (bytes - tcp-retx-bytes) * 100 / bytes |
| tcp-retx-bytes |
Count of TCP retransmitted bytes |
| tcp-retx-pkts |
Count of TCP retransmitted packets, excluding tossed packets |
| tcp-retx-pkts% |
TCP retransmit rate — the percentage of packets seen by PacketWise
that were retransmitted
Formula: tcp-retx-pkts / tcp-data-pkts |
Table of Link Variables
PacketWise stores link data for the variables in this table.
| Link Variables |
Description |
| accelerated-bytes |
Number of bytes that were accelerated |
| acceleration-mode |
Indicates whether acceleration is on, off, or has
changed during the interval (1=off, 2=changed, 3=on) |
| hostdb-alloc-fails |
Number of times the system was unable to allocate
resources for host database flows during the measurement interval.
* |
| ipdg-alloc-fails |
Number of times the system was unable to allocate
resources to classify/control a non-TCP flow during the measurement
interval. * |
| link-size-bps |
Configured link size |
| redirect-pkts |
Number of packets redirected from the PacketShaper to a cache device (such as the iShared appliance) |
| returned-pkts |
Number of packets that are returned from a cache device (such as the iShared appliance) that need to be de-encapsulated and sent to their destination |
| pkt-size-histogram |
Histogram of number of packets received in different
size buckets. The buckets of packet sizes (in bytes) include: [0-63],
[64-127], [128-255], [256-511], [512-1023], [1024-1517], [>=1518] |
| rx-errors |
Count of received packets that were dropped due to hardware errors;
measures data at the NIC level
Note: Because this variable measures data at the NIC level,
there will be minimal data recorded in the outbound direction when
watch mode is enabled (the only outbound traffic will be PacketShaper
management). |
| rx-no-buffers |
Count of packets that were dropped due to unavailable buffers;
measures data at the NIC level. *
Note: Because this variable measures data at the NIC level,
there will be minimal data recorded in the outbound direction when
watch mode is enabled (the only outbound traffic will be PacketShaper
management). |
| rx-pkts-dropped |
Count of received packets that were dropped due to:
• no buffers available (also counted separately in rx-no-buffers
variable)
• no inbound/outbound connections available
• random dropping
This variable measures data at the NIC level.
Note: Because this variable measures data at the NIC level,
there will be minimal data recorded in the outbound direction when
watch mode is enabled (the only outbound traffic will be PacketShaper
management). |
| rx-pkts-dropped% |
Percentage of received packets that were dropped
Formula: rx-pkts-dropped / total-rx-pkts |
| shaping-mode |
Indicates whether traffic shaping is on, off, or has changed during
the interval (1=off, 2=changed, 3=on)
Note: This variable is considered experimental and may be
changed or removed in the future. |
| tcp-alloc-fails |
Number of times the system was unable to allocate
resources to classify/control a TCP flow during the measurement interval.
* |
| total-passthru-bytes |
Total number of bytes in the total-passthru-pkts |
| total-passthru-pkts |
Total number of packets that were passed through
the unit without being controlled |
| total-rx-bytes |
Total number of bytes received, not including bytes
that were dropped |
| total-rx-pkts |
Total number of packets received, not including
packets that were dropped |
| total-sameside-pkts |
Number of packets that were classified into the SameSide traffic
class (a class that measures traffic not destined for the WAN) regardless
of whether the packets are dropped or passed through
Note: This variable is considered experimental and may be
removed in the future. |
| total-tx-bytes |
Total number of bytes transmitted, not including
bytes that were dropped |
| total-tx-pkts |
Total number of packets transmitted, not including
packets that were dropped |
| tx-errors |
Count of transmitted packets that were dropped
due to hardware errors. * |
| tx-pkts-dropped |
Count of transmitted packets that were dropped
due to:
• no route available
• I/O errors
• no buffer available |
| unsolicited-icmp |
Detect denial-of-service attacks; detects an ICMP reply when there
was no matching request |
* Frequent non-zero values indicate the system is running at peak
capacity.
Table of Partition Variables
PacketWise stores partition data for the variables in this table.
| Partition Variables |
Description |
| commitment-failure% |
Percentage of one-minute intervals in which packets were dropped
when the partition size was less than 6% below the reserved partition
size. For example, if packets were dropped in 6 different one-minute
intervals during an hour period, the commitment-failure% would be
10 (percent). If no packets were dropped during the interval, the
value would be 0. If packets were dropped in a one-minute interval,
the value would be 100. |
| dynamic-cap-count |
Number of flows that were given admission control because the dynamic
partition already had the maximum number of users |
| dynamic-live-user |
Peak number of live users in a dynamic partition during the measurement
interval (live user is defined as a user who is actively using
the application and won’t be usurped if a new user wants a subpartition
and none is available) |
dynamic-no-
partition-count |
Number of flows that were denied because the PacketShaper had
reached its maximum number of partitions and new subpartitions could
not be created |
| late-drop-bytes |
Number of bytes in a partition that were dropped because they had
not yet been transmitted after eight seconds or more of delay queuing |
| late-drop-pkts |
Number of packets in a partition that were dropped because they
had not yet been transmitted after eight seconds or more of delay
queuing |
lowest-fully-
satisfied-priority |
Lowest priority traffic level that could get all the bandwidth
it requested
Note: Because this variable is sampled multiple times in
the interval, the lowest fully satisified priority may be one that
occurred only once.
|
| partition-burst-limit-bps |
Configured partition burst limit; same as the partition size if
the partition is not burstable |
| partition-over-limit-msecs |
Cumulative time during which allocated bandwidth exceeded the partition
minimum size, in milliseconds |
| partition-over-limit-secs |
Cumulative time during which allocated bandwidth exceeded the partition
minimum size, in seconds |
| partition-over-limit-time% |
Percentage of time that allocated bandwidth exceeded the partition
minimum size
Formula: partition-over-limit-msecs / sample-interval-msecs |
| partition-size-bps |
Configured partition minimum size |
| pvc-avg-bps |
Rate in bps based on the pvc-bytes over the interval |
| pvc-avg-fps |
Frames-per-second rate |
| pvc-bytes |
Number of bytes sent or received on the permanent virtual circuit
(PVC) |
| pvc-ecn-frames |
Number of frames that had FECN/BECN (Forward/Backward Explicit
Congestion Notification) set |
| pvc-ecn-frames% |
Percentage of frames with FECN/BECN set |
| pvc-frames |
Number of frames in the interval |
| pvc-target-bps |
Average target rate for the PVC |
| sched-drop-bytes |
Number of bytes in a partition that were dropped due to insufficient
buffer space |
| sched-drop-pkts |
Number of packets in a partition that were dropped due to insufficient
buffer space |
Table of Class Variables
PacketWise stores class measurement data for the variables in this table.
| Class Variables |
Description |
| app-availability% |
Percentage of time the service is available
The app-availability% variable has the following characteristics:
- It measures the availability when the application is actively
in use and bases it upon transactions that are attempted and responded
to. The variable does not perform a periodic ping to a server.
- It reflects the availability of the application based upon
the success of connection attempts to a server. Therefore, if
no connection attempts are made over a period of time (for example,
at 2am), the availability will be reported as 100%, even if the
application was temporarily down for maintenance.
- At the smallest measurement granularity — one minute —
the app-availability% variable reports either 0% or 100%. If all
connection attempts within the one-minute period fail, then availability
is considered to be 0%. If any connections succeeded or there
were no connection attempts, availability is considered to be
100%. Essentially, at one-minute intervals, the app-availability%
variable offers a simple flag to indicate whether any connection
attempts were successful.
- At intervals longer than one minute, the percentage is averaged
across the interval.
|
| avg-round-trip-time |
Average number of milliseconds a packet takes to go from client
to server and back again |
| class-hits |
Number of times flows match the class; class hits occur only at
the beginning of a flow or session |
| client-flood-block |
Number of flows that were blocked due to a client (flow initiator)
exceeding the flow limit rate specified in the policy
flowlimit command |
| conn-speed-hist |
Speed histogram over well-known speeds. This histogram provides
a profile of use, for example, for dial-up modems versus high-speed
modems. Use this data to identify which levels are overburdened. These
speeds represent the PacketWise-detected speeds per flow. This data
is recorded only for FTP and HTTP connections. All other classes report
zeroes. |
| icmp-attempt-count |
Count of attempted ICMP pings |
| icmp-rtt-count |
Count of received icmp-echo_reply messages |
| icmp-rtt-msecs-sum |
Sum of ICMP ping round-trip times in milliseconds |
| icmp-rtt-msecs-avg |
Average ICMP ping round-trip time in milliseconds |
| icmp-success-rate% |
ICMP success rate (icmp-rtt-count/icmp-attempt-count) |
| drop-bytes% |
Percentage of transmitted bytes dropped due to bandwidth saturation
Formula: no-bandwidth-drop-bytes / total-bytes-tx |
| license-overflows |
Count of the number of flows that would exceed the licenses |
| licenses-peak |
The largest value of licenses (flows) issued during the interval.
Note that the value of the licenses-peak variable is reported as 0
if no licenses were issued during the interval regardless of how many
are in use from prior periods. |
| licenses-total |
The value of the class
licenses limit at the end of the interval |
| mos-x10 |
Mean Opinion Score, a value derived from the R-Factor per ITU‑T Recommendation G.10, measures VoIP call quality (see rfactor). Packeteer measures MOS using a scale of 10-50. To convert to a standard MOS score (which uses a scale of 1-5), divide the Packeteer MOS value by 10.
|
| network-delay-avg |
Average number of milliseconds the class' transaction packets
spent in transit |
| network-delay-histogram |
Histogram of the number of milliseconds the class' transaction
packets spent in transit
Note: Each histogram contains a table of data where the
index represents the lower bound of a time range in milliseconds:
[25000], [10000], [5000], [2500], [1000], [750], [500], [250], [100],
[75], [50], [25], [10], [0]. Table cells contain the number of transactions
whose delay time fell within the range represented by the index. |
| network-delay-median |
Median number of milliseconds the class' transaction packets spent
in transit (half the class' network delays were shorter, and half
were longer) |
| network-delay-msec |
The sum of the network delays of all transactions in the specified
interval, measured in milliseconds. (Network delay is the time a transaction
spends in transit.) This variable is useful for calculating weighted
averages across multiple intervals. |
| no-bandwidth-drop-bytes |
Number of bytes dropped due to bandwidth saturation |
| normalized-network-delay-avg |
Transaction delay in the network, normalized by transaction size;
measured in milliseconds per kilobyte
Note: This variable is experimental and may be removed in the
future. |
| peak-ipdg-conns |
Peak number of concurrent non-TCP flows |
| pkt-exchange-time-avg |
Average packet exchange time
Formula: pkt-exchange-time / pkt-exchange-time-samples |
| pkt-exchange-time-msecs |
Interval between when a data packet leaves the
PacketShaper and its ACK arrives (ACKs to SYN packets are not counted) |
| pkt-exchange-time-samples |
Number of pkt-exchange-time samples that were counted
|
| policy-hits |
Number of times the policy has been enforced |
| rfactor |
Rating Factor. A value derived from metrics such as latency, jitter, and packet loss per ITU‑T Recommendation G.107, the R-Factor value helps you quickly assess the quality-of-experience for VoIP calls on your network. Typical scores range from 50 (bad) to 90 (excellent). |
| round-trip-time-msecs |
The sum of the round-trip-times (RTT) of all transactions in the
specified interval, measured in milliseconds. (RTT is the time a packet
takes to go from client to server and back again.) Note that this
measurement is taken once per transaction (not once per packet). This
variable is useful for calculating weighted averages across multiple
intervals. |
| rtp-latency-count |
Number of RTP latency samples used to compute latency |
| rtp-latency-sum |
Sum of measured RTP latency |
| rtp-latency-avg |
Average RTP latency in milliseconds; for instance, if the sum of measured latency is 500 ms and the number of latency samples is 25, then the average latency is 20 ms.
Formula: rtp-latency-sum / rtp-latency-count
Note: Measurement data for rtp-latency-avg isn't calculated until you turn on the Enable Latency Calculations system variable. |
| rtp-expected-pkts |
Expected number of RTP packets |
| rtp-lost-pkts |
Number of lost RTP packets |
| rtp-lost-pkt% |
Percentage of expected RTP packets that have been lost |
| rtp-jitter-count |
Number of jitter samples used to compute jitter |
| rtp-jitter-sum |
Sum of measured jitter |
| rtp-jitter-avg |
Average jitter in milliseconds; for instance, if the sum of measured jitter is 100 milliseconds and the number of jitter samples is 25, then the average jitter is 4 milliseconds.
Formula: rtp-jitter-sum / rtp-jitter-count |
| server-delay-avg |
Average number of milliseconds required for servers to process
the class' transaction requests. The time starts when the server
has received all required request packets and ends when the server
issues the first packet of the response. |
| server-delay-histogram |
Histogram of the number of milliseconds required for servers to
process requests for the class' transactions. For example, these measurements
could monitor HTTP response for an Internet server, if you set up
a host-specific class for the server. |
| server-delay-median |
Median number of milliseconds required for servers to process the
class' transaction requests (half the class' server delays were
shorter, and half were longer) |
| server-delay-msec |
The sum of the server delays of all transactions in the specified
interval, measured in milliseconds. (Server delay is the time required
for servers to process the class' transaction requests.) This variable
is useful for calculating weighted averages across multiple intervals. |
| server-flood-block |
Number of flows that were blocked due to a server (flow destination)
exceeding the flow limit rate specified in the policy
flowlimit command |
| service-level% |
Percentage of transactions that satisfied their performance requirement
(as defined by the total-delay-threshold variable); the percentage
of good or sufficiently speedy transactions. |
| service-level-errors |
Number of one-minute intervals that do not have the required percentage
of speedy transactions, as defined by service-level-threshold% |
| service-level-threshold% |
Maximum percentage of slow transactions (as defined by the total-delay-threshold)
that each one-minute interval can have and still be considered an
acceptable interval |
| slow-transactions |
Number of slow transactions (as defined by the total-delay-threshold
variable) |
| total-bytes-tx |
Number of bytes transmitted in a traffic class
Note that this variable differs from the link variable total-tx-bytes, which measures the total bytes transmitted on a link. |
| total-delay-avg |
Average number of milliseconds to complete the class' transactions;
includes network delay and server delay. |
| total-delay-histogram |
Histogram of the number of milliseconds required to complete the
class' transactions |
| total-delay-median |
Median number of milliseconds required to complete the class'
transactions (half the class' total delays were shorter, and half
were longer) |
| total-delay-msec |
The sum of the delays of all transactions in the specified interval,
measured in milliseconds. (Total delay is the time required to complete
a transaction; includes network and server delay.) This variable
is useful for calculating weighted averages across multiple intervals. |
| total-delay-threshold |
Number of milliseconds that constitutes "too slow" for a total
delay (time required for a transaction to complete) |
| total-trans |
Number of transactions (request-response pairs) |
| trans-bytes |
Transaction size for TCP-based applications |
| trans-bytes-avg |
Average number of bytes per transaction
Formula: trans-bytes / total-trans |
| web-response-2XX |
Number of HTTP response messages with 2XX success
codes — for example, 200 OK and 201 Created; this
variable tracks responses of any class that has web traffic, including
synthetic transactions |
| web-response-3XX |
Number of HTTP response messages with 3XX redirection
codes — for example, 301 Moved and 302 Found; this
variable tracks responses of any class that has web traffic, including
synthetic transactions |
| web-response-4XX |
Number of HTTP response messages with 4XX client
error codes — for example, 400 Bad Request and 404
Not found; this variable tracks responses of any class that has
web traffic, including synthetic transactions |
| web-response-5XX |
Number of HTTP response messages with 5XX server
error codes — for example, 501 Not implemented and 502
Timed out; this variable tracks responses of any class that has
web traffic, including synthetic transactions |
|
