Hello, I am running a probe [0] at my home and I have 240Mbit connection from my provider. When I look at the graphs, this probe is basically sitting idle doing nothing! I can spare 20Mbit/s easily for measurements to keep the usage. However, I do not know how to achieve that. Is it possible to attract more traffic to my probe? I'm sure there is some algorithm involved to distribute the work so if you can please get me some more work, I'd be pleased [obviously more credits to spare :)] [0] https://atlas.ripe.net/probes/31252/ Regards, -- . 73! DE TA1AET http://erenturkay.com/
In general I believe the RIPE Atlas network is heavily underutilized. I can confirm the low traffic too: Now there are a few reasons that this could happen, but in general I believe the entire network is like this, and not just specific probes.. The reasons can be: i) People just don’t have anything to measure, so they don’t send new measurements ii) People think credits are actually super important, and want to keep them iii) People believe current measurements are too expensive, so theoretically if everything costed 10x less, there would be 10x the measurements. Now personally I think that it’s just (i), but I can’t actually really tell. Changing the available bandwidth from the RIPE Atlas page seems to not affect the bandwidth used. On the other hand, somehow the traffic is relatively constant at 7-8 Kb/s, and then moved up to like 8-9 Kb/s.. I do not think RIPE Atlas has 100% evenly distributed measurements, so there’s possibly some mechanism to distribute the load uniformly?
On 25 Aug 2018, at 15:02, Eren Türkay <turkay.eren@gmail.com> wrote:
Hello,
I am running a probe [0] at my home and I have 240Mbit connection from my provider. When I look at the graphs, this probe is basically sitting idle doing nothing! I can spare 20Mbit/s easily for measurements to keep the usage. However, I do not know how to achieve that.
Is it possible to attract more traffic to my probe? I'm sure there is some algorithm involved to distribute the work so if you can please get me some more work, I'd be pleased [obviously more credits to spare :)]
[0] https://atlas.ripe.net/probes/31252/ <https://atlas.ripe.net/probes/31252/>
Regards,
-- . 73! DE TA1AET http://erenturkay.com/ <http://erenturkay.com/>
Hi! Eren, thank you for additional probe in Atlas network, I will for sure use it :D I can add 2 reason that stops some people from using your probe. 1. Some people (like me) for various valid reasons cannot use general Atlas setting for probe allocation like "give me all probes from Ireland". Such people use more sophisticated algorithms to use probes that they are interested in and they refresh their private probe list every X days/weeks. Your probe looks quite fresh so probably you have to wait for cronjob to update researchers' probes lists ;) 2. Your probe is pretty fresh so it doesn't have system tags like 'IPv4 Stable 1d' or 'IPv4 Stable 90d'. I know people that uses only probes with 'IPv4 Stable 30d' or better. Their argument is that they want to use only stable probes which can give them comparable samples over time. And this is a good approach if you have big measurement network but I personally think that Atlas is still too small in many countries so using only 'IPv4 Stable 30d' or better can heavily reduce number of samples that you get. Probes come and go so we should use whatever is available if our objective is to make large scale measurement globally. In short, don't worry. With time you should see more measurements. Remember also that for some researches your location and provider that you use can be just no interesting. Regards, Grzegorz From: Antonios Chariton <daknob.mac@gmail.com> Date: Saturday 2018-08-25 at 15:17 To: Eren Türkay <turkay.eren@gmail.com> Cc: "ripe-atlas@ripe.net" <ripe-atlas@ripe.net> Subject: Re: [atlas] Getting More Traffic for Probe In general I believe the RIPE Atlas network is heavily underutilized. I can confirm the low traffic too: [http://44.128.63.10/rrdash/rrd.php?from=-2400000&to=-30&graph=atlas] Now there are a few reasons that this could happen, but in general I believe the entire network is like this, and not just specific probes.. The reasons can be: i) People just don’t have anything to measure, so they don’t send new measurements ii) People think credits are actually super important, and want to keep them iii) People believe current measurements are too expensive, so theoretically if everything costed 10x less, there would be 10x the measurements. Now personally I think that it’s just (i), but I can’t actually really tell. Changing the available bandwidth from the RIPE Atlas page seems to not affect the bandwidth used. On the other hand, somehow the traffic is relatively constant at 7-8 Kb/s, and then moved up to like 8-9 Kb/s.. I do not think RIPE Atlas has 100% evenly distributed measurements, so there’s possibly some mechanism to distribute the load uniformly? On 25 Aug 2018, at 15:02, Eren Türkay <turkay.eren@gmail.com<mailto:turkay.eren@gmail.com>> wrote: Hello, I am running a probe [0] at my home and I have 240Mbit connection from my provider. When I look at the graphs, this probe is basically sitting idle doing nothing! I can spare 20Mbit/s easily for measurements to keep the usage. However, I do not know how to achieve that. Is it possible to attract more traffic to my probe? I'm sure there is some algorithm involved to distribute the work so if you can please get me some more work, I'd be pleased [obviously more credits to spare :)] [0] https://atlas.ripe.net/probes/31252/<https://urldefense.proofpoint.com/v2/url?u=https-3A__atlas.ripe.net_probes_31252_&d=DwMFaQ&c=96ZbZZcaMF4w0F4jpN6LZg&r=XA7UqyUiWhQ8cJzGZlnLA89Ync-_75lLOvUCSOji-4w&m=zjksFQ-AyE4oar1Ati4lZfBH7MTj2aNLYMnL2gSYpQY&s=4OsoLgAXbXU39_lAz6txbRp_3Nrnn4lr72XNL0GnKmY&e=> Regards, -- . 73! DE TA1AET http://erenturkay.com/<https://urldefense.proofpoint.com/v2/url?u=http-3A__erenturkay.com_&d=DwMFaQ&c=96ZbZZcaMF4w0F4jpN6LZg&r=XA7UqyUiWhQ8cJzGZlnLA89Ync-_75lLOvUCSOji-4w&m=zjksFQ-AyE4oar1Ati4lZfBH7MTj2aNLYMnL2gSYpQY&s=q1kSA66x25A1-Qmr7-ZZKvEMdWVZ2z7K2oP9E_yMHCo&e=>
You maybe right. I will wait a couple of weeks until my probe is tagged stable. It is fairly stable now, I fixed the boot problem and installed new OpenWRT router which works great for me. If somebodycan manually tag and shift traffic, it will handle without problem. Nevertheless, I will wait and report back. Thanks for suggestions! I hope this probe will be better utilized. As I said, it can easily handle 20Mbit/s (even more). On Sat, Aug 25, 2018 at 2:17 PM Antonios Chariton <daknob.mac@gmail.com> wrote:
In general I believe the RIPE Atlas network is heavily underutilized. I can confirm the low traffic too:
Now there are a few reasons that this could happen, but in general I believe the entire network is like this, and not just specific probes..
The reasons can be:
i) People just don’t have anything to measure, so they don’t send new measurements ii) People think credits are actually super important, and want to keep them iii) People believe current measurements are too expensive, so theoretically if everything costed 10x less, there would be 10x the measurements.
Now personally I think that it’s just (i), but I can’t actually really tell. Changing the available bandwidth from the RIPE Atlas page seems to not affect the bandwidth used. On the other hand, somehow the traffic is relatively constant at 7-8 Kb/s, and then moved up to like 8-9 Kb/s.. I do not think RIPE Atlas has 100% evenly distributed measurements, so there’s possibly some mechanism to distribute the load uniformly?
On 25 Aug 2018, at 15:02, Eren Türkay <turkay.eren@gmail.com> wrote:
Hello,
I am running a probe [0] at my home and I have 240Mbit connection from my provider. When I look at the graphs, this probe is basically sitting idle doing nothing! I can spare 20Mbit/s easily for measurements to keep the usage. However, I do not know how to achieve that.
Is it possible to attract more traffic to my probe? I'm sure there is some algorithm involved to distribute the work so if you can please get me some more work, I'd be pleased [obviously more credits to spare :)]
[0] https://atlas.ripe.net/probes/31252/
Regards,
-- . 73! DE TA1AET http://erenturkay.com/
-- . 73! DE TA1AET http://erenturkay.com/
I can obly guess, but 20 Mbps for ICMP, DNS or HTTP probess will result in CPU saturation on that box. For us it’s much important is stable Probe anybody can use anytime. — Evgeniy Sudyr On Sat 25. Aug 2018 at 21:01, Eren Türkay <turkay.eren@gmail.com> wrote:
You maybe right. I will wait a couple of weeks until my probe is tagged stable. It is fairly stable now, I fixed the boot problem and installed new OpenWRT router which works great for me. If somebodycan manually tag and shift traffic, it will handle without problem. Nevertheless, I will wait and report back.
Thanks for suggestions! I hope this probe will be better utilized. As I said, it can easily handle 20Mbit/s (even more).
On Sat, Aug 25, 2018 at 2:17 PM Antonios Chariton <daknob.mac@gmail.com> wrote:
In general I believe the RIPE Atlas network is heavily underutilized. I can confirm the low traffic too:
Now there are a few reasons that this could happen, but in general I believe the entire network is like this, and not just specific probes..
The reasons can be:
i) People just don’t have anything to measure, so they don’t send new measurements ii) People think credits are actually super important, and want to keep them iii) People believe current measurements are too expensive, so theoretically if everything costed 10x less, there would be 10x the measurements.
Now personally I think that it’s just (i), but I can’t actually really tell. Changing the available bandwidth from the RIPE Atlas page seems to not affect the bandwidth used. On the other hand, somehow the traffic is relatively constant at 7-8 Kb/s, and then moved up to like 8-9 Kb/s.. I do not think RIPE Atlas has 100% evenly distributed measurements, so there’s possibly some mechanism to distribute the load uniformly?
On 25 Aug 2018, at 15:02, Eren Türkay <turkay.eren@gmail.com> wrote:
Hello,
I am running a probe [0] at my home and I have 240Mbit connection from my provider. When I look at the graphs, this probe is basically sitting idle doing nothing! I can spare 20Mbit/s easily for measurements to keep the usage. However, I do not know how to achieve that.
Is it possible to attract more traffic to my probe? I'm sure there is some algorithm involved to distribute the work so if you can please get me some more work, I'd be pleased [obviously more credits to spare :)]
[0] https://atlas.ripe.net/probes/31252/
Regards,
-- . 73! DE TA1AET http://erenturkay.com/
-- . 73! DE TA1AET http://erenturkay.com/
-- -- With regards, Eugene Sudyr
Yeah, Atlas wasn't designed to generate heavy traffic (it's not iperf ;)). Check page with most busy probes and you will see what you can expect. https://atlas.ripe.net/get-involved/community/ Good think about 200Mbps link is that have a lot of headroom so we can be quite sure that we will see not spoiled results. Regards, Grzegorz From: Evgeniy Sudyr <eject.in.ua@gmail.com> Date: Saturday 2018-08-25 at 21:15 To: Eren Türkay <turkay.eren@gmail.com> Cc: "ripe-atlas@ripe.net" <ripe-atlas@ripe.net> Subject: Re: [atlas] Getting More Traffic for Probe I can obly guess, but 20 Mbps for ICMP, DNS or HTTP probess will result in CPU saturation on that box. For us it’s much important is stable Probe anybody can use anytime. — Evgeniy Sudyr On Sat 25. Aug 2018 at 21:01, Eren Türkay <turkay.eren@gmail.com<mailto:turkay.eren@gmail.com>> wrote: You maybe right. I will wait a couple of weeks until my probe is tagged stable. It is fairly stable now, I fixed the boot problem and installed new OpenWRT router which works great for me. If somebodycan manually tag and shift traffic, it will handle without problem. Nevertheless, I will wait and report back. Thanks for suggestions! I hope this probe will be better utilized. As I said, it can easily handle 20Mbit/s (even more). On Sat, Aug 25, 2018 at 2:17 PM Antonios Chariton <daknob.mac@gmail.com<mailto:daknob.mac@gmail.com>> wrote: In general I believe the RIPE Atlas network is heavily underutilized. I can confirm the low traffic too: [http://44.128.63.10/rrdash/rrd.php?from=-2400000&to=-30&graph=atlas] Now there are a few reasons that this could happen, but in general I believe the entire network is like this, and not just specific probes.. The reasons can be: i) People just don’t have anything to measure, so they don’t send new measurements ii) People think credits are actually super important, and want to keep them iii) People believe current measurements are too expensive, so theoretically if everything costed 10x less, there would be 10x the measurements. Now personally I think that it’s just (i), but I can’t actually really tell. Changing the available bandwidth from the RIPE Atlas page seems to not affect the bandwidth used. On the other hand, somehow the traffic is relatively constant at 7-8 Kb/s, and then moved up to like 8-9 Kb/s.. I do not think RIPE Atlas has 100% evenly distributed measurements, so there’s possibly some mechanism to distribute the load uniformly? On 25 Aug 2018, at 15:02, Eren Türkay <turkay.eren@gmail.com<mailto:turkay.eren@gmail.com>> wrote: Hello, I am running a probe [0] at my home and I have 240Mbit connection from my provider. When I look at the graphs, this probe is basically sitting idle doing nothing! I can spare 20Mbit/s easily for measurements to keep the usage. However, I do not know how to achieve that. Is it possible to attract more traffic to my probe? I'm sure there is some algorithm involved to distribute the work so if you can please get me some more work, I'd be pleased [obviously more credits to spare :)] [0] https://atlas.ripe.net/probes/31252/<https://urldefense.proofpoint.com/v2/url?u=https-3A__atlas.ripe.net_probes_31252_&d=DwMFaQ&c=96ZbZZcaMF4w0F4jpN6LZg&r=XA7UqyUiWhQ8cJzGZlnLA89Ync-_75lLOvUCSOji-4w&m=nvCxALZWQvmIjviuXu9dk4piTp7e6X9tc1CF3QC9xGw&s=fYTjrx-H57wxUCloDxvDP6IjG1b4e6RpIWjw5Zh02kE&e=> Regards, -- . 73! DE TA1AET http://erenturkay.com/<https://urldefense.proofpoint.com/v2/url?u=http-3A__erenturkay.com_&d=DwMFaQ&c=96ZbZZcaMF4w0F4jpN6LZg&r=XA7UqyUiWhQ8cJzGZlnLA89Ync-_75lLOvUCSOji-4w&m=nvCxALZWQvmIjviuXu9dk4piTp7e6X9tc1CF3QC9xGw&s=DG9kjuBv7wIxWd_ifwn-PeBWCfK5F3jnIO08UFGhr_M&e=> -- . 73! DE TA1AET http://erenturkay.com/<https://urldefense.proofpoint.com/v2/url?u=http-3A__erenturkay.com_&d=DwMFaQ&c=96ZbZZcaMF4w0F4jpN6LZg&r=XA7UqyUiWhQ8cJzGZlnLA89Ync-_75lLOvUCSOji-4w&m=nvCxALZWQvmIjviuXu9dk4piTp7e6X9tc1CF3QC9xGw&s=DG9kjuBv7wIxWd_ifwn-PeBWCfK5F3jnIO08UFGhr_M&e=> -- -- With regards, Eugene Sudyr
participants (4)
-
Antonios Chariton -
Eren Türkay -
Evgeniy Sudyr -
Ponikierski, Grzegorz