- Nebula Measurement Results Calendar Week 32 - 2024
The following results show measurement data that were collected in calendar week 32 in 2024 from 2024-08-05 to 2024-08-12.
- Number of crawls
336 - Number of visits
25,860,738Visiting a peer means dialing or connecting to it. Every time the crawler or monitoring process tries to dial or connect to a peer we consider this as visiting it. Regardless of errors that may occur.
- Number of unique peer IDs visited
70,958 - Number of unique peer IDs discovered in the DHT
70,538 - Number of unique IP addresses found
137,085
Timestamps are in UTC if not mentioned otherwise.
Newly discovered agent versions:
github.com/JackalLabs/sequoia@14e37f7f5(2024-08-05 07:21:08)kubo/0.30.0-dev/2f78519ef-dirty(2024-08-05 18:21:00)helia/4.2.5 libp2p/1.8.3 UserAgent=v18.16.0(2024-08-05 19:51:00)kubo/0.30.0-dev/749a61bae-dirty/docker(2024-08-06 11:21:09)helia/4.2.4 libp2p/1.8.1 UserAgent=v22.4.1(2024-08-06 13:20:49)kubo/0.30.0-dev/40cf63087(2024-08-06 15:21:07)kubo/0.30.0-dev/ffab7b2/docker(2024-08-06 23:21:44)kubo/0.30.0-dev/fc31f9c/docker(2024-08-07 00:21:04)kubo/0.30.0-dev/84389e8/docker(2024-08-07 03:51:01)kubo/0.30.0-dev/a4a05a1/docker(2024-08-07 04:20:51)kubo/0.30.0-dev/a4a05a184-dirty(2024-08-07 13:21:40)server@(2024-08-08 04:21:37)helia/4.2.5 libp2p/1.8.3 UserAgent=v18.19.1(2024-08-08 05:51:01)helia/4.2.5 libp2p/1.8.3 UserAgent=v21.7.2(2024-08-08 09:21:15)helia/4.2.4 libp2p/1.8.2 UserAgent=v20.14.0(2024-08-08 10:50:56)helia/4.2.1 libp2p/1.6.0 UserAgent=v20.15.1(2024-08-08 16:51:21)github.com/JackalLabs/sequoia@15842eeda-dirty(2024-08-08 19:50:52)github.com/JackalLabs/sequoia@f38fc173c-dirty(2024-08-08 23:51:04)kubo/0.30.0-dev/a4a05a1(2024-08-09 01:20:51)github.com/JackalLabs/sequoia@5074ad959-dirty(2024-08-09 13:21:02)github.com/functionland/go-fula@b390bf0b2(2024-08-10 01:20:49)js-libp2p/0.45.9 UserAgent=v22.5.1(2024-08-10 03:21:39)
Agent versions that were found to support at least one storm specific protocol:
go-ipfs/0.8.0/48f94e2storm
Newly discovered protocols:
/flamingo/channel/1.0.0(2024-08-06 17:50:52)/flamingo/channel/segment/1.0.0(2024-08-06 17:50:52)/flamingo/cache/1.0.0(2024-08-06 17:51:48)/yang/net/v1(2024-08-08 04:21:37)/my-echo/0.1(2024-08-10 03:21:39)
A "rotating node" is a node (as identified by its IP address) that was found to host multiple peer IDs.
| IP-Address | Country | Unique Peer IDs | Agent Versions | Datacenter IP |
|---|---|---|---|---|
51.75.78.61 |
DE | 190 | ['edgevpn'] | True |
51.15.133.64 |
IN | 114 | ['kubo/0.22.0-dev/c95f9e1-dirty'] | True |
51.159.150.159 |
FR | 90 | ['kubo/0.22.0-dev/c95f9e1-dirty'] | True |
152.81.47.227 |
FR | 90 | ['hydra-booster/0.7.4'] | False |
51.158.232.73 |
FR | 87 | ['kubo/0.22.0-dev/c95f9e1-dirty'] | True |
151.115.53.194 |
PL | 80 | ['kubo/0.22.0-dev/c95f9e1-dirty'] | True |
151.115.53.194 |
PL | 80 | ['kubo/0.22.0-dev/c95f9e1-dirty'] | True |
51.15.86.171 |
FR | 79 | ['kubo/0.22.0-dev/c95f9e1-dirty'] | True |
51.15.247.127 |
FR | 78 | ['kubo/0.22.0-dev/c95f9e1-dirty'] | True |
51.15.90.150 |
FR | 78 | ['kubo/0.22.0-dev/c95f9e1-dirty'] | True |
Only the top 10 kubo versions appear in the right graph (due to lack of colors) based on the average count in the time interval. The 0.8.x versions do not contain disguised storm peers.
storm* are go-ipfs/0.8.0/48f94e2 peers that support at least one storm specific protocol.
You can find the most up-to-date graph on probelab.io/ipfskpi.
In the specified time interval from 2024-08-05 to 2024-08-12 we visited `` unique peer IDs.
All peer IDs fall into one of the following classifications:
| Classification | Description |
|---|---|
offline |
A peer that was never seen online during the measurement period (always offline) but found in the DHT |
dangling |
A peer that was seen going offline and online multiple times during the measurement period |
oneoff |
A peer that was seen coming online and then going offline only once during the measurement period |
online |
A peer that was not seen offline at all during the measurement period (always online) |
left |
A peer that was online at the beginning of the measurement period, did go offline and didn't come back online |
entered |
A peer that was offline at the beginning of the measurement period but appeared within and didn't go offline since then |
Only the top 10 kubo versions appear in the right graph (due to lack of colors) based on the average count in the time interval. The 0.8.x versions do not contain disguised storm peers. This graph also excludes peers that were online the whole time. You can read this graph as: if I see a peer joining the network, what's the likelihood for it to stay X hours in the network.
storm* are go-ipfs/0.8.0/48f94e2 peers that support at least one storm specific protocol.
Only the top 10 kubo versions appear in the right graph (due to lack of colors) based on the average count in the time interval. The 0.8.x versions do not contain disguised storm peers.
storm* are go-ipfs/0.8.0/48f94e2 peers that support at least one storm specific protocol.
Includes all peers that the crawler was able to connect to at least once: dangling, online, oneoff, entered. Hence, the total number of peers is lower as the graph excludes offline and left peers (see classification).
storm shows the go-ipfs/0.8.0/48f94e2 peers that support at least one storm specific protocol.
The classifications are documented here.
storm* are go-ipfs/0.8.0/48f94e2 peers that support at least one storm specific protocol.
This graph shows all IP addresses that we found from 2024-08-05 to 2024-08-12 in the DHT and their geolocation distribution by country.
The classifications are documented here. The number in parentheses in the graph titles show the number of unique peer IDs that went into the specific subgraph.
storm* are go-ipfs/0.8.0/48f94e2 peers that support at least one storm specific protocol.
This graph shows all IP addresses that we found from 2024-08-05 to 2024-08-12 in the DHT and their datacenter association.
The classifications are documented here. Note that the x-axes are different.
The number in parentheses in the graph titles show the number of unique peer IDs that went into the specific subgraph.
storm* are go-ipfs/0.8.0/48f94e2 peers that support at least one storm specific protocol.
For a description of our measurement methodology check out this repository.
The time it took to receive the first byte of the first response (that was not a redirect). The large number in each tile is the time in seconds. The number at the very bottom of the graph shows the sample size that went into each subplot/website. Note: the color scales are different in each graph.
The number above each bar shows the sample size that went into the calculation.
The following graph shows the daily error rate in accessing these website.
We are running lean libp2p peers that just support the Kademlia DHT protocol in six different AWS regions. Each peer takes turns to publish the provider record for a CID of random data. All other peers are then instructed to lookup that CID. "Looking up" here means finding the provider record. So the numbers below don't show the actual content retrieval times (which would depend on file sizes) but instead the DHT performance. The peers run go-libp2p-kad-dht version v0.21.1 + default configurations.
Code can be found here: dennis-tra/parsec (we plan to move this to our ProbeLab organization)
The number in parenthesis is the number of publications/retrievals for that particular region that went into the calculation.
The number in the box is the number of publications/retrievals for that particular region that went into the calculation.
The box extends from the first quartile (Q1) to the third quartile (Q3) of the data, with a line at the median. The whiskers extend from the box by 1.5x the inter-quartile range (IQR). Flier points are those past the end of the whiskers. See https://en.wikipedia.org/wiki/Box_plot for reference.
The number in the box is the number of publications that went into the calculation of the box.
The box extends from the first quartile (Q1) to the third quartile (Q3) of the data, with a line at the median. The whiskers extend from the box by 1.5x the inter-quartile range (IQR). Flier points are those past the end of the whiskers. See https://en.wikipedia.org/wiki/Box_plot for reference.
The number in the box is the number of publications/retrievals that went into the calculation of the box.
The box extends from the first quartile (Q1) to the third quartile (Q3) of the data, with a line at the median. The whiskers extend from the box by 1.5x the inter-quartile range (IQR). Flier points are those past the end of the whiskers. See https://en.wikipedia.org/wiki/Box_plot for reference.
| Classification | Description |
|---|---|
offline |
A peer that was never seen online during the measurement period (always offline) but found in the DHT |
dangling |
A peer that was seen going offline and online multiple times during the measurement period |
oneoff |
A peer that was seen coming online and then going offline only once during the measurement period |
online |
A peer that was not seen offline at all during the measurement period (always online) |
left |
A peer that was online at the beginning of the measurement period, did go offline and didn't come back online |
entered |
A peer that was offline at the beginning of the measurement period but appeared within and didn't go offline since then |
The following protocol strings are unique for storm nodes according to this Bitdefender paper:
/sreque/*/shsk/*/sfst/*/sbst/*/sbpcp/*/sbptp/*/strelayp/*