Permanent private measurement
Hi Atlas folks, I'd like to use my probe to monitor the status of my home Internet connection by continuously pinging my ISP's network (which is not the first or second hop in traceroute, but something like the sixth) and perhaps www.google.com. Is there a way to do this easily using the existing UI and tools? I couldn't find any existing measurements for this. I can look at the built-in v4 and v6 measurements for K-root, which is normally very close to me, but that's not 100% representative of the status of my Internet connection because it also measures the peering exchange link between my ISP and K-root. Cheers, Lorenzo
Hi Lorenzo,
On Oct 7, 2015, at 2:01 PM, Lorenzo Colitti <lorenzo@google.com> wrote:
Hi Atlas folks,
I'd like to use my probe to monitor the status of my home Internet connection by continuously pinging my ISP's network (which is not the first or second hop in traceroute, but something like the sixth) and perhaps www.google.com <http://www.google.com/>. Is there a way to do this easily using the existing UI and tools? You can do a traceroute and configure the TTL (Maximum hops) to 6.
I couldn't find any existing measurements for this. I can look at the built-in v4 and v6 measurements for K-root, which is normally very close to me, but that's not 100% representative of the status of my Internet connection because it also measures the peering exchange link between my ISP and K-root.
Cheers, Lorenzo
Cristel
On Oct 7, 2015, at 2:17 PM, Pelsser Cristel <cristel@iij.ad.jp> wrote:
Hi Lorenzo,
On Oct 7, 2015, at 2:01 PM, Lorenzo Colitti <lorenzo@google.com <mailto:lorenzo@google.com>> wrote:
Hi Atlas folks,
I'd like to use my probe to monitor the status of my home Internet connection by continuously pinging my ISP's network (which is not the first or second hop in traceroute, but something like the sixth) and perhaps www.google.com <http://www.google.com/>. Is there a way to do this easily using the existing UI and tools? You can do a traceroute and configure the TTL (Maximum hops) to 6. The first hop also needs to be set to 6.
I couldn't find any existing measurements for this. I can look at the built-in v4 and v6 measurements for K-root, which is normally very close to me, but that's not 100% representative of the status of my Internet connection because it also measures the peering exchange link between my ISP and K-root.
Cheers, Lorenzo
Cristel
On Wed, Oct 7, 2015 at 2:17 PM, Pelsser Cristel <cristel@iij.ad.jp> wrote:
I'd like to use my probe to monitor the status of my home Internet connection by continuously pinging my ISP's network (which is not the first or second hop in traceroute, but something like the sixth) and perhaps www.google.com. Is there a way to do this easily using the existing UI and tools?
You can do a traceroute and configure the TTL (Maximum hops) to 6.
Thanks. In the end I hardcoded the IPv6 address for the ISP gateway, since I haven't seen it change often. I suppose I'll live to regret this. The main part of my question was "how do I set up a measurement on my own probe that runs forever" and I think the answer is "just use a user measurement". I ended up setting up three ping measurements, each one sending 4 packets each spaced by 15000ms, every minute, because that's what will fit in my credit income. It's strange that I can't send one ping every 15 seconds, but there you are. Also, the visualization is not as nice as the built-in measurements (why?) but it seems to do what I want.
It's strange that I can't send one ping every 15 seconds, but there you are. Also, the visualization is not as nice as the built-in measurements (why?) but it seems to do what I want.
If you could be a bit more specific (which visualisation?) then we can check what we can do about it. Generally speaking, the visualisations for built-ins need a somewhat different strategy than for UDMs, since we always expect a lot of results there. Also, the newly introduced latencymon is probably your best choice for this kind of viz. See https://labs.ripe.net/Members/massimo_candela/new-ripe-atlas-tool-latencymon Cheers, Robert
On Wed, Oct 7, 2015 at 4:01 PM, Robert Kisteleki <robert@ripe.net> wrote:
It's strange that I can't send one ping every 15 seconds, but there you are. Also, the visualization is not as nice as the built-in measurements (why?) but it seems to do what I want.
If you could be a bit more specific (which visualisation?) then we can check what we can do about it.
What I am really trying to do is what the built-in measurements do - run continuous measurements, from my probe only, to relevant targets on the Internet. I want to do this so I can measure the performance and reliability of my last-mile connectivity without factoring in any peering latency or anything else. I think this is what the built-in "traceroute second hop" measurement is intended to do, but that doesn't work for me because I (and the rest of my ISP's customers) am a variable number of hops away from the Internet. This is because the NTT NGN, unlike most unbundled access networks which use tunneling technologies such as L2TP or PPPoE, uses native IPv6. So the number of hops between customer and ISP network depends on NGN topology and routing. I worked around this by setting up a UDM that sends pings, but that's a bit of a hacky substitute, and as you say, the UDM visualizations aren't really geared to this use case. Also, the newly introduced latencymon is probably your best choice for this
kind of viz. See
https://labs.ripe.net/Members/massimo_candela/new-ripe-atlas-tool-latencymon
Latencymon sort of works (as does seismograph), but it's definitely not as nice as the built-in measurement UI.
Morning Lorenzo, If I've understood your problem, you have already found a way to make the measurements from your probe but you don't like how the results are plotted. There is a GIT repository from Atlas with examples about how to get the results in json format and print them using rrdtool. You can even use a nice JavaScript library or import them to nagios or zabbix. Regards, Daniel Am 08.10.2015 3:56 vorm. schrieb "Lorenzo Colitti" <lorenzo@google.com>:
On Wed, Oct 7, 2015 at 4:01 PM, Robert Kisteleki <robert@ripe.net> wrote:
It's strange that I can't send one ping every 15 seconds, but there you are. Also, the visualization is not as nice as the built-in measurements (why?) but it seems to do what I want.
If you could be a bit more specific (which visualisation?) then we can check what we can do about it.
What I am really trying to do is what the built-in measurements do - run continuous measurements, from my probe only, to relevant targets on the Internet. I want to do this so I can measure the performance and reliability of my last-mile connectivity without factoring in any peering latency or anything else.
I think this is what the built-in "traceroute second hop" measurement is intended to do, but that doesn't work for me because I (and the rest of my ISP's customers) am a variable number of hops away from the Internet. This is because the NTT NGN, unlike most unbundled access networks which use tunneling technologies such as L2TP or PPPoE, uses native IPv6. So the number of hops between customer and ISP network depends on NGN topology and routing.
I worked around this by setting up a UDM that sends pings, but that's a bit of a hacky substitute, and as you say, the UDM visualizations aren't really geared to this use case.
Also, the newly introduced latencymon is probably your best choice for this
kind of viz. See
https://labs.ripe.net/Members/massimo_candela/new-ripe-atlas-tool-latencymon
Latencymon sort of works (as does seismograph), but it's definitely not as nice as the built-in measurement UI.
participants (4)
-
Daniel Gomez -
Lorenzo Colitti -
Pelsser Cristel -
Robert Kisteleki