Test Traffic Measurements FAQs
- What is the purpose of the Test Traffic Measurements
service (TTM)?
- What are TTM Test Boxes?
- How does TTM work?
- What is IP Packet Delay Variation (IPDV), also
called "jitter"?
- What are "valid packets"?
- Why is it important plotting the 2.5th and 97.5th
percentiles for delay measurements?
- What does "RMS" mean?
- How much traffic do Test Boxes generate?
- We received and installed our Test Box, but
why does it not give a login prompt?
- How can the IP address of a Test Box be changed?
- What kind of access do Test Boxes need to and
from the outside world?
- We changed our filter rules, what should we
check?
- What are "monitoring ranges"?
- Why do I receive alarm messages?
- We received an invoice but something is wrong
- Why did you include VAT on our invoice?
What is the purpose of the Test Traffic Measurements
service (TTM)?
TTM's main purpose is to provide standardized metrics for one-way delay
and one-way packet loss between measurement devices in a format, which
is easily understood by users. One-way measurements are vital, since more
and more of Internet routing is becoming asymetrical, i.e. packets use
different paths between two nodes depending on the direction of traffic.
The TTM project complies with standards that are maintained and developed
by the IETF's IP Performance Metrics Working Group (RFC's 2330, 2678 through
2681).
What are TTM Test Boxes?
Test Boxes are measurement devices with GPS (Global Positioning System)
antennas connected to them for synchronization and increased time accuracy
(approx. 100 ns). They generate test-traffic, receive test-traffic and
perform trace routes to determine path vectors between boxes. The traffic
generated resembles "real" traffic as much as possible to avoid
preferential treatment by network devices in the path between two measurement
points.
How does TTM work?
One-way measurements imply that information for calculating metrics and
plotting data will scattered over any two measurement devices. Consequently,
central servers at the RIPE NCC must regularly collect the information,
make calculations and generate the plots in order to present the "Big
Picture". They also manage the TTM operations by sending control
messages to individual Test Boxes and acting as a central repository for
measurement software.
What is IP Packet Delay Variation (IPDV),
also called "jitter"?
IPDV measures the variation in delay of uni-directional, consecutive packets
(packet 1 and 2, 2 and 3 etc.) which flow between between two hosts over
an IP path. Low IPDV is especially important for applications requiring
timely delivery of packets, e.g. multimedia applications, VoIP, video
etc. The maximum delay variation is useful for determining the optimal
buffer sizes for such applications.
See URL for more information:
http://www.ietf.org/internet-drafts/draft-ietf-ippm-ipdv-10.txt
What are "valid packets"?
Valid packets are packets that have good timestamps and can therefore
be used for TTM statistics. Bad packets are usally sent during unreliable
GPS conditions, meaning that their timestamps are inaccurate. Bad packets
are discarded when calculating TTM metrics.
Why is it important plotting the 2.5th
and 97.5th percentiles for delay measurements?
The 2.5th and 97.5th percentiles represent the one-way delay between two
Test Boxes in a best-case and worst-case scenario respectively, just as
the median (50th percentile) would indicate a normal value between two
measurement points.
In many cases, the so-called standard deviation can be used to describe
this as well. However, for some of the delay distributions, it can be
argued that the value cannot be properly calculated. Percentiles do not
suffer from this problem.
What does "RMS" mean?
Physical scientists often use the term "root-mean-square" as
a synonym for "standard deviation" when referring to the square
root of the mean squared deviation of a signal from a given baseline.
In our case, high values could be caused either by frequent path vector
changes, unstable/congested links or a combination of both. Low values
would naturally imply the opposite. Of course, there may be other interpretations
too.
It can be argued that the RMS or standard deviation should not be used
to describe the data considering the nature of the delay distributions.
How much traffic do Test Boxes generate?
By default, approximately 300 kB per day to every destination Test Box.
The defaults can be changed on series C/D.
Defaults per Destination Test Box:
Test-Traffic Packet Payload Size: 100 bytes
Test-Traffic Packet Tx Rate: 2 packets/min
We received and installed our Test Box,
but why does it not give a login prompt?
It occasionally happens that a connector came lose during transport. Please,
open the cover of the Test Box ( 4 screws) and press all connectors firmly
into their sockets, in particular the flat ribbon cables for the disks
and the cards on the motherboard. Also make sure that the disk trays are
in the locked position.
If the problem persists, then swap the disks and try again. Please contact
us if still unresolved.
How can the IP address of a Test Box
be changed?
Please send
us the new IP address, netmask & default router at least two working
days before the renumbering should be effective.
TT-ops will notify you when the change has been configured. When your
new addressing schema is ready, press the reset button to reboot the machine.
After rebooting, ping the new IP address to make sure that the box is
there. Data-taking will restart automatically.
What kind of access do Test Boxes need
to and from the outside world?
We require the following (see also RIPE 179).
- ICMP
- TCP
- Outgoing mail (SMTP)
- Incoming SSH (from 193.0.0.0/23)
- UDP
- Incoming/outgoing NTP (port 123)
- Incoming/outgoing DNS (local resolver)
- Incoming/outgoing TestTraffic (ports 1024 and higher).
We changed our filter rules, what should
we check?
Telnet to the box and check if you still see lines starting with RCDP
there. Then contact TT-Ops
and inform them about the change.
The Test Box operators at the NCC should check:
- RCDP data files continue to collect data
- Other boxes continue to receive data from this box
- Operational mails still arrive at tt-ops@ripe.net
- The local named can resolve DNS queries (try "host tt01 127.0.0.1"
and "host tt01 193.0.0.198")
What are "monitoring ranges"?
Customers can specify ranges of IP addresses which may access the Test
Box. In other words, "monitoring ranges" can be defined as a
kind of "permit access-list" for allowing source hosts access
to the measurement device. TTM operators are naturally included in the
monitoring ranges. Please beware that hosts not within these ranges will
have access denied.
Why do I receive alarm messages?
If the the delay between two Test Boxes changes significantly, then an
email will be sent to the operators of those boxes.
The alarm uses a "long-term average, short-term average" (LTA-STA)
algorithm, which compares certain percentiles and medians for a 30-day
interval with dito for a 30-minute interval. More specifically, if the
5th percentile for the shorter interval exceeds the 95th percentile for
the longer one, i.e. 95% of recent observations are higher (or worse)
than 95% of values registered over the past 30 days, then an alarm will
be sent to the specified operators every 30 minutes until the condition
disappears. The LTA-STA algorithm will adjust to daily descepancies in
traffic patterns by deviding the day into four 6-hour-intervals (0.00-6.00,
6.00-12.00 etc) and then comparing them separately to avoid false alarms.
- Click here
for more information about Test Box alarms.
We received an invoice but something
is wrong.
For technical questions about invoices, such as:
We need more information before we can pay,
The invoice was sent to the wrong person,
There is a mistake in adding up the various amounts,
Can you send us a new copy of the invoice,
please contact us by email at billing@ripe.net.
If you believe that we made a mistake with the contents of the invoice
(for example: we charged you for equipment that was not delivered), please
write to tt-ops@ripe.net.
In all cases, please mention the number of your test-box and the number
of the invoice.
Why did you include VAT on our invoice?
Even though most customers in the EU are do not have to pay VAT, local
regulations force us to charge VAT unless they have provided us with a
valid EU VAT number. To avoid being charged VAT at all, please mail
us your EU VAT number.
If your organization is exempt from VAT and your invoice mentions VAT,
then send us your VAT number and ask for a new invoice.
In all cases, please mention both the number of your test-box
and the invoice number.
|