An In-depth Picture of the Current Ethereum Network by MigaLabs | Devcon SEA

[Music] thank you uh thank you for being here so I'm Leo from migal laabs um so I want to start this talk with a quote from one of my um favorite scientists uh computer science uh dictra and he says computer science is no more about computers than astronomy is about telescopes and I think this is a great great example and talking about as omy um I want to show this picture is a picture of the center of the um Galaxy and actually this two pictures look completely different it's actually exactly the same spot is the center of the Milky Way and this is how it looks with us two different types of telescopes the one on the um that you see on the left is taken by a radio telescope and the other one is used by a normal kind of telescope so this is what scientists do they create tools to look at reality with different uh from different perspectives and this is the work what we do uh uh IM meals so similarly in the ethereum space you sometimes look at data from different perspective and this is an example that I want to show what you see on the left side is um the client Distribution on the consensus layer that we observe from crowling the network and connecting with different Bon nodes U on the n on the peer-to-peer layer right and when you look at the right side uh you have a similar picture that shows um how this look when you look at uh the creation of blocks using a tool that is called blog print which we actually try to improve and and manage to improve the accuracy using a different machine learning uh technology we actually you can discard the QR code if you look at the pap if you want to look at the paper but what I want to show here is that if you really look at the numbers there are some differences between one and the other because uh what really happens is that sometimes we have many many more validators in one beon node and then when you took of these two pictures you will see different realities in the same way you do when you see with two different types of telescopes and this is what is really important that to really understand um what is going on looking at reality from different perspectives so here for example if you look at the green te has like 12% of the bcoin nodes but in reality is creating 23% of the of the blocks in this is because apparently they are focusing on no with really har um High number of of validators uh if you look at the execution layer diversity this is what we can see doing some graffiti analysis about 59% is G uh 36% is nether mind if you look at the same picture one year ago you will see a completely different thing so we have evolved but we still have to do some work here uh of course not all the blocks in the system have graffitis that can that allow us to do this work only about 133% of this of the of the blocks have a graffiti that allow us to analyze this so uh this is important actually I want to say for you guys that are running notes in the in the audience uh we have seen the quite significant Improvement of of useful graffitis for in the last six months uh so that's good uh but for those that are running noes please uh try to help on this so that we scientists can look at the data and and have an idea of how things are evolving uh now if you look at uh if you want to look at things like uh ISP decentralization and entity decentralization well we have a lot of data also in our dashboard the QR code is uh is a link to monitor edwi where we publish all these data um the figure on the right shows the different I mean the different entities that are uh stake in ethereum uh those that are in blue are entities that have less than 10,000 validators the ones in green have between 10,000 and to 50,000 validators in yellow you see from 50,000 to 100,000 validators and in red you have coinbase with over 120,000 validators okay so what we want to see in this picture is less red and more blue okay that's what this message should be about and if you look at uh the isps about 15% of the of the nodes are running on Amazon Cloud that's not a huge number people are usually believing that is much much larger than that so this is not a bad bad picture but it will be great to see more home nodes is taken in the network so this is that still work that um that we have to do now um this is the geographical distribution of the ethereum nodes again what we want to see in this picture is less red and more more blue about 30% of the nodes are running in the US and there are plenty of data centers all around the world so maybe if you are running noes in clouds maybe we can move those to other places so that we have a much better geographical distribution as well anybody knows how many nodes run in Thailand in theum network any expectation 23 23 nodes running uh in Thailand in the in the tum Network okay basically what I wanted to show is that um we wanted to um we are working on developing multiple tools to have different visions of the theum network and if you're interested in our work please come to our booth next to the classroom uh C and or you can call me uh and and ask me questions offline thank you so much thank you Leo now time for k um as the question in I have a question you mentioned that in Thailand there are 20 three nodes I'm curious to know which country has the highest number of nodes yeah so that would be the United States with uh 35% of the of the nodes been run in the United States yes okay second place and second second place will be the US H sorry the uh Germany Germany has about 16% of the nodes and the third place is France okay interesting um that means there's room for other countries to speed up on the amount of nodes um the first question how do you get to this data again so we have uh multiple tools uh one of the tools that we have is a crowler that is connecting to the ethereum nodes in the consensus layer and we are developing another crowler that is connecting to the nodes in the execution layer in addition to that we use other tools like chatu which is developed uh by the devop steam the panda of steam of the EF uh that is also Gathering a lot of data and we also use that data we also developing another tool that is called goth um that is actually uh one of the tools that we have used to do this analysis on on on um the beon uh sorry the the noes the blocks and uh we also have another tool that is uh is called Ed poer which is actually to analyze the this decentralization of entities in um in in staking okay then the next one re claim they have roughly 5% market share but your data shows 0.16 market share are they lying so uh we have seen uh what I what I show here was just the graffiti that is shown in the block and as I explained in the in the slide uh only 133% of the blocks have a useful graffiti so what I sh what I what I show here the figure was about the data that we collected from 133% of the blocks that has that are stuck with a graffiti that we can understand so that means it's possible that they have 5% and it's and at the same time they figure that I show is correct okay thank you so much Leo time is up but um very insightful um talk I think you're still around if people have more questions um talk to Leo after thank you so much Leo thank you