Amp Enterprise Grade Blockchain Data that Scales I Kevin Jones and Marcus Rein

Awesome. Uh, can you guys hear me? Okay. Yeah. I feel like I have to like lean over here like, "Hey, what's up?"

Uh, hello everybody. Uh, my name is Kevin Jones. I am a developer relations engineer. I'm also joined with Marcus here. Uh, we are co-presenting this.

Um, and very I don't know what happened there. Anyways, okay. I'm trying not to touch it. Um, yeah. So, we're going to be talking today about AMP.

Uh, very exciting. This was just released today, so you guys are getting very, very fresh content. Um, yeah. So, let's just dive right into it. I think we want to like kind of just talk about like what the bounties are and uh then we're going to do a walk through of AMP and we're going to actually show you how to use it.

Uh first of all, if you want to build on AMP, there are three different categories uh or sorry, if you want to build on the graph, there's three different categories. The first one that we're looking for people to do is the best use of AMP data sets. So basically that means you just need to use AMP in your hackathon project. Okay. Uh and we're going to go much into detail on what AMP is, but there's first, second, and third place.

Then we also have two other categories. the best use of the graph for AI and this is including the subgraph MCP server and the token API MCP server and using that in your AI agent or AI use case. And then we have another for best use of sub subgraph advanced features. These are all listed on the ETH global website on what those features are. So just be sure to make sure that you know what those are.

Uh the the repo that Marcus is going to be going through is here. Uh this is the AMP hackathon repo. Huge shout out to Marcus. He's been like really like grinding away on this and he's created a really great experience for you as a hacker to get started quickly using this repo. So we're excited to show that off.

You can just go to the edge node GitHub and yeah so what is AMP? So first of all um I think it makes sense just to talk briefly about why we created AMP. Um one of the things is we wanted to kind of do a almost like a reboot right a restart at how we imagine getting access to blockchain data. So you guys know we have subgraphs which is like our flagship product. Uh AMP is kind of a reboot on that that allows you to get access to uh data through a SQL endpoint.

So it's a database for blockchain data. Really really powerful and very exciting. Uh AMP is a in a developer preview. So this is a brand new product. Uh but to be honest with you, it's really really impressive and very very flexible.

Uh and we're looking for lots of feed feedback. So if you have feedback about the experience, please let us know. We want your your um uh your input on how we can make it better. So please let us know. Uh also AMP is all about local development.

We're all about uh making a local development and that is exactly what Marcus is going to do today. He's going to show you how easily you can do everything locally and spin up your project and get the access to your data using AMP uh in that environment. And on top of that, again, it's powered by SQL. SQL is like super powerful. And there's a lot of really cool uh features in AMP that make it easy to query and get access to raw blockchain data or derived data that you can uh create with with AMP as well.

You can also do streaming or you can do batch query transformations. So, this is very nice. You can actually just get like full batches of particular uh information from AMP or you can stream data directly uh across the wire into your application which makes it very powerful and we also automatically handle reorg reorgs on the chain uh so you don't need to worry about that. We also support multi-chain and multicontract. This is one of like the I guess things with subg graphs that you could not do automatically out of the box and is a huge uh powerful point of what you can do with AMP.

And also the data is verifiable because we have access to we're actually running um uh directly connected to uh the either the RPC or via fire hose and we're verifying that data as we put it into parquet files uh in the data set. Uh and we also have AMP powered subgraphs that are coming. So if you are used to using subgraphs and you still want the graphql API that you're used to uh but you still want to uh power them with AMP in the speed that you get with uh with the data processing that is also coming soon as well. Uh this makes again all together the main thing about AMP is it's very very performant and highly scalable and very flexible for for building a project uh for data. Uh so just some metrics.

So we have access to the chain head at about 100 milliseconds uh give or take. You can uh decode 4 million events per second with AMP and you can query all of the block hashes since Genesis on mainnet in 5 seconds. So that kind of gives you a brief clash of how powerful it is. And on top of that, we took and we crunched the entire uh transactions on all of mainet and put them either to two categories that has value or does not. And we were able to do that in 35 seconds.

So we were able to like split and put them into two buckets like 1.5 billion rows of data. So super powerful. So let's do it. Uh let's bring up Marcus.

Yes. [applause]

Uh thank you Kevin. Hi everyone. I'm Marcus. I'm senior developer relations at Edge and Node. Let's get building and let's talk about databases.

If anyone here has interacted with the graph before, you think indexing, indexing, indexing. This is database, database, database. It's important to have that distinction because if you want to interact with a database, that means filtering. So if you before wanted to use subgraphs or substreams, how you would get a specific contract addresses data, you would do it differently than this. So, as you're going in through the process of learning about AMP, just keep on going back to if you want specifics, learn SQL if you don't know it.

And if you know it, you're going to be in a good place. Okay. The next thing, if you want to have interesting transformations, you could absolutely do that on your data. You have the raw data, it's in the paret files, you could run transformations on that using SQL and create new tables. That's awesome.

And if you want to use something else such as a TypeScript UDF to have more customization and more power, you'd prefer to use TypeScript, there's pre-baked ones in the repository right now. And they're really nice for decoding. They're really nice for doing more advanced transformations. Eventually, there will be more transformations available by you guys because custom transformations will soon be available. Okay, so let's go into actually the hackathon.

And I think the mo the most important concept of this presentation and what we're trying to get across is data sets. So there's two different types of tables that can be created by AMP. There's raw and derived. Raw is just that. It's raw blockchain data which can be stored locally.

It could be on prem. It could be in the cloud. It's a database. It could be stored wherever you want. And you could run transformations however you'd like on top of that raw data.

If you want to then potentially change up your strategy and have your data sets settled with transformed data, pre-transformed data, that's a derived table. Okay? You're essentially deriving more interesting tables specific to your use case by using a derived table and derived data sets. Is that clear? We have two different types of data sets, two different types of tables.

Yes. Okay. So, now you could all win money at the hackathon. That's that's good. Okay.

So, um let's let's get into building a little bit. Um, now I'm going to go ahead and say I I've I've been to many ETH global hackathons. I am really enjoy them. They are great and also sometimes there's no internet which is okay. Uh, the intent of this is to say, hey, local is possible.

So, what we're seeing right now is completely local. We're indexing locally. We can actually interact with smart contracts. We can go ahead and use SQL locally. This is a local environment that we can have.

And then we can create our database as such. So we have the logs going right here. And if you want to look, this is the fun part I think. We go into AMP and look at the data. We have our raw data here.

Blocks, logs, transactions, raw data. And we have our derived data. I've named it counter. We have active blocks, decremented, incremented. And if we just interact with our front end, all right, we see logs are changing.

That's cool. We actually have the parquet files materializing in real time into our database. As I'm clicking here, it is populating our database, streaming right in right away. And that is a Postgress database that contains parquet files, portable. It's a database.

Do whatever you want with it. And that's really powerful. So, um, I want to unleash this database concept to you guys. Uh, for many years, we've been focusing on indexing, indexing, indexing. You guys don't want indexing.

You guys want data. You guys want data in a database. And that's the most important thing that we all want to convey with this. Being able to get data from chain, either it be a local chain or from onchain into these data sets, be it raw or derived, is very fast. And once you have that database, you could do whatever you want with it using SQL.

So, um, there's a few different toolings that we have available right now for a better development experience. This is just the UI that we have for the little baby app. But if you want, we also have studio available, which I think is running. Let me see. Yes, we have studio running.

Uh, let's just shut it down because this logo uh pudio. There we go. And it should pull open. There it is. And this is the local development experience.

Um, being able to run queries from a UI is quite nice. You could also take a look at the events. You can see the different sources. Just a different development environment. Look for look out for this to be more extended in the future.

Not just testing queries, but being able to interact with data sets. Uh, anything you'd want to do such as composing data sets. Let's say you have a data set of amazingly derived tables and you said, "Hey, let's plug that into another data set." That's all possible at least locally. We're going to get to that point.

It is al also possible with our published data sets. Right now I'm in a local environment. I can build. That's great. The uh published environment is just that where you have let's see if I can find it.

A published environment means come on. Nope. There it is. Uh the playground. So whenever you run an AMP instance, you are essentially saying I want to have AMP and the data sets potentially coming all locally bringing it from Anvil or from onchain and you want to house that data that's totally fine.

There's also hosted data. If you don't don't want to worry about gigabytes and terabytes of data and having it in tables, there's a hosted uh uh service provided by Edge and Node that you could just go to, this is online right now, and build off of that. So you don't even have to run AMP at all. You could just go ahead and use SQL to query blockchain data. Uh one last unlock uh that I want to convey before time for questions.

Uh transformations are really important. Uh a lot of times data analysts uh app developers want data in a specific way and that might change dayto-day. And with uh this you're going to be able to use SQL to essentially say at query time I want it to be arranged in this table filtered this way joined this way from this chain this chain that smart contract that smart contract and bring it all to me through one query. That's really powerful. Okay, I want to convey how powerful that is because this is multi-chain.

This is multicontract. Being able to hit multiple chains and multiple contracts all at once with one query udfs included. That's super powerful. And so, please get in here. Break everything that you see.

[snorts] This is uh developer preview, which I know I've got some serious senior developers in here. Break it. Find things that are uncomfortable. Talk to Kevin and talk to me. Uh I want to make this better so we can make web 3 and crypto better.

So, thank you guys for your time. Uh if there's any questions, uh Kevin and I will be here. And is there anything else that I missed here? Kevin, are we good to go?

Oh, yeah. Yeah, we could do that. Uh, the readme is quite inclusive. I was working on this a lot. Um, but essentially it has a quick start.

This comes with the repo. In fact, I'll put the QR in there. It's uh just install, just up, just dev. It's really quick to get going locally. It also has a pathway.

If you don't even want to start locally and you just want to go to the drive data sets that are online that are already hosted all the way from Genesis block already available for you, you can do that too through a basically start with this readme. gives you a pathway to build locally, move to remote, or just start in remote and get building. That's what the prizes are for. Um, okay, cool. Uh, let me pull up the QR code for, uh, that repository.

There it is. And, uh, are there any questions? I think we have time for questions. Yes.

Yes. Okay. Questions.

Okay. Hey. Oh, we got a got a question.

Hey.

Hey. How do you handle reorganizations of the blockchain?

Yes, it handles reorganizations automatically. Our senior engineers are about a 100 meters behind you and they'll be able to go into great detail on exactly how that's handled.

Okay. But if it's automatic, I don't really care about it.

I'm sorry, I can't hear you.

If it's automatic, I don't really care about it. So it's you already answered my question. Thank you.

Perfect. Okay, there you go.

Yeah.

Uh I don't need to launch a subgraph. I can query directly.

You do not need to launch a great question. Subgraphs and AMP are different. They do connect if you want them to. One of our engineers just pushed something that uh right that just came in. So it should connect today.

I'm I I'm not 100% sure on that, but it should. It's on its way. But yes, they're two separate things. Um, if you build locally, you're just working on an Anvil instance, or it could be hard hat, too. As long as you connect to the RPC locally, you're good to go.

And then from there, that's one environment. You could also work on data sets that are published that have all of the data from we have arbitrum, base, Ethereum, uh, and base Solia all the way from Genesis block and you can just send SQL to that location too. And so that's that's does that answer your question?

Yeah. So basically, I deploy a contract on base and I can query it directly without

bingo. If you deploy it on base and just hit the endpoint that's on the remote data set, it has it right. Cool. Thank you. That's it.

Sorry, can you repeat which chains are supported?

Great question. So we have uh currently four chains supported, one test net. We have arbitum, we have bass, we have Ethereum, all mainet and then we have base also supported.

Oh, arbitum.

Arbitum. Okay,

good to know. Thank you. Uh, arbitrum is not supported yet because it's not synced fully. Uh, it'll be available by Sunday.

Okay, cool.

Yeah.

This is developer preview, mind you. We would we are looking forward um there's please go to the read me. There's a few different um uh caveats with streaming transformations. There's a few different caveats right now. This is developer preview.

Um, so I want to get feedback from developers of what you want to build and what we're stopping you from building. If you have an idea and you're like, "Dang it, I want that." Please talk to us because we want that, too. We want to unlock that stuff for you. We're aware of things that we are aware of.

We're trying to unlock, but of course, we don't have uh uh omnipotent vision. So, please come to us with your vision. We'll help you out.

Okay.

Okay. All right. Thank you guys for your time. [applause]