Hey, guys!
Who can share any hands on experience with The Graph, Subquery, Subsquids, Kenshi, Nakji in real building?
it seems there is a lot of solutions, but really interesting to know the differences or what is better for what
Jun 9, 2023, 12:46 PM
Kenshi is great for webhook
Jun 13, 2023, 8:32 PM
I have only worked with the Graph and Dune
I interviewed for a job in Dune at some point and got some insights from the founders. They didnt build it for scalability to begin with an I personally think that using SQL to query blockchain data seems like not the most intuative interface. They have later been able to address some of the scalability issues afaik.
I have used the graph in hoppy projects. I like the way to query and it seems more intuative and better for blockchain projects. That being said It feels very centralised ( I know they later have done some iniatives to address it so It may not be a problem in the lon haul ).
-
One question that I have been thinking is why this sort of stuff isnt implemented more closely to the actual blockchian infrastructure of the individual projects. Since nodes out there are already processing transactions why not allow them to indesx as well and serve the indexed data, should they want to ?
I interviewed for a job in Dune at some point and got some insights from the founders. They didnt build it for scalability to begin with an I personally think that using SQL to query blockchain data seems like not the most intuative interface. They have later been able to address some of the scalability issues afaik.
I have used the graph in hoppy projects. I like the way to query and it seems more intuative and better for blockchain projects. That being said It feels very centralised ( I know they later have done some iniatives to address it so It may not be a problem in the lon haul ).
-
One question that I have been thinking is why this sort of stuff isnt implemented more closely to the actual blockchian infrastructure of the individual projects. Since nodes out there are already processing transactions why not allow them to indesx as well and serve the indexed data, should they want to ?
Jun 14, 2023, 2:50 AM
Generally, The Graph and Dune are really different things
1. Dune is a commercial centralized platform for analysts to easily create dashboards using the same skills that they had before (SQL). And as it is usually happens with analytical tools it has a significant latency (~minutes) that makes it impossible to use in real-time applications.
2. The Graph (which is similar to SubQuery, SubSquids, etc.) is a decentralized network which makes users (here indexers) run the nodes (graph-node here) and these nodes running only those subgraphs that has been signalled by curators. And it is a huge layer so that it is not really possible just to unite it with the ordinary blockchain nodes. This layer also has its own consensys and tokenomics that is also really different and solves its own problem - how to access indexer smart contract data in a decentralized way for anyone
And here goes an important thing. What we are actually comparinng with Dune is not The Graph decentralized network, but its centralized implementation called "hosted service" that has been made by The Graph to allow developers run/debug their subgraphs in a sandbox. And this became really popular solution when the indexing engine is taken from The Graph and hosted without any decentralized stuff. But to compare it with any other production grade solution (like Dune) one have to utilize not The Graph's sandbox, but also a production grade tool like we do at chainstack subgraphs, because there are significant differences in speed of sync (~3x), query latency (~2x) and reliability.
and always there is available option to host your own (graph-node + database + archive node) because it is open source and under MIT licence, but there is no such an option for closed source Dune.
1. Dune is a commercial centralized platform for analysts to easily create dashboards using the same skills that they had before (SQL). And as it is usually happens with analytical tools it has a significant latency (~minutes) that makes it impossible to use in real-time applications.
2. The Graph (which is similar to SubQuery, SubSquids, etc.) is a decentralized network which makes users (here indexers) run the nodes (graph-node here) and these nodes running only those subgraphs that has been signalled by curators. And it is a huge layer so that it is not really possible just to unite it with the ordinary blockchain nodes. This layer also has its own consensys and tokenomics that is also really different and solves its own problem - how to access indexer smart contract data in a decentralized way for anyone
And here goes an important thing. What we are actually comparinng with Dune is not The Graph decentralized network, but its centralized implementation called "hosted service" that has been made by The Graph to allow developers run/debug their subgraphs in a sandbox. And this became really popular solution when the indexing engine is taken from The Graph and hosted without any decentralized stuff. But to compare it with any other production grade solution (like Dune) one have to utilize not The Graph's sandbox, but also a production grade tool like we do at chainstack subgraphs, because there are significant differences in speed of sync (~3x), query latency (~2x) and reliability.
and always there is available option to host your own (graph-node + database + archive node) because it is open source and under MIT licence, but there is no such an option for closed source Dune.
Jun 14, 2023, 5:34 AM