Jerry did a good one on this today as well, I'm sure you saw, I appreciate your added insight, what we need is Dominic Williams to address this and let us know in laymans terms, why this is occurring and what are possible solutions if any?
Adding more nodes to subnets will do nothing, it will be only be worse because update call has to reach consensus in the subnet which means every update call runs on all nodes anyway, so subnet is as strong as the one node strong. We need to increase node power or make switching canisters between subnets automatically, so you can scale your app running on multiple subnets without even knowing. Or we need to make sharding in the subnet it self to not run everything on all nodes, but that may be same as running canisters on multiple subnets. Other thing that it seems that what overflows subnet is many small update calls because of additional work needed for nodes to handle those update calls, so maybe we need to increase base update call price, so developers will do less of small calls but more of bigger calls.
Awesome. That's one of the reasons I made the video; to help get people talking about it. And yes, obviously I am a content creator as well, so having something to talk about that might get people to watch is also part of the equation.
PLEASE COMMENT ON MY COMMENT A COUPLE THREADS UP... 😊We need to have separate chains for Trading (Memes creating calls) and possibly separate chains for token creation and actual subnet projects cuz memes only need social engagement, but if they need subnets, then separate Bob from icp-tv (a project i will be building in the future, but in this current structure i will have to expand my business and deploy on utopia which would suck as it should definitely be a community thing.)😢
They deployed a fix to one subnet yesterday and it seems to have fixed the issue. Im hoping they prioritise scaling a lot more now and forget about AI for a while
@@ASHdemoREEL yes, but only on one subnet. It looks like the fix went well and i can confirm the canisters me and my team deployed are not suffering the issue anymore. Over time the other subnets will get the same fix. However, DF still need to prioritise heart beats and the scheduler since the fix might not be the silver boot
Thankyou @@TheOmfg02 for your clarification, much appreciated. as an ICP developer, do you believe this is a simple solve for ICP? or could is this a dealbreaker? and could be the leak before the damn fully crumbles?
@@ASHdemoREEL there are a lot of options open to DF to improving the scalability issues. They just haven’t prioritised it for a long time. I understand they wanted to move on to different features and progress with their roadmap. This isn’t a deal breaker and im still confident they can improve things a lot. It’s not simple in terms of the time and complexity needed to create the solution but the solutions are possible with their current team quality, size and budget so i have good faith in them. It’s just a shame they didn’t do more testing to earlier to find this potential issue but again, it’s hard to create a testing environment that would expose the scaling issue
Thanks so much @@TheOmfg02 for your knowledgeable opinion. For current investors, is this fix solveable for the foreseeable current bullrun, or its a fix for the next 5-8 years?
Thanks again for a great and informative video, I guess I was a little disappointed when I found out about the issues however I do think that in order for ICP to evolve into greatness, there has to be some growing pains...
@@jjwe2002 Dom is not concerned with retail and their faux pas. this is not a Dfinity problem, it's clear the devs are not versed on the Motoko language, yet. in the meantime let Dom focus on his target client base,.. enterprise, NGO's, military, and nation state contracts. gl
This was brought up in 2021 and they said they were fixing it shortly, now a dev is saying that they shouldn’t build with a 1 canister per user model. Hate to be that guy but that’s the whole point of ICP. How many times has Dominic said each person will have their own AI running as a canister smart contract? I truly hope this gets resolved soon
A very similar thing happened to EOS. "Mining" projects hit it to extract resources and the chain was somewhat unusable until they came up with solutions. Growing pains and it certainly seems like these apps that did onboard a lot of users had a huge bill. Tough to come up with an appropriate funding model and a setup that can scale properly.
Developer may have had a poor design from the start. Canister is a Project (not a User). My worry NNS may not be aware of these shortfalls by bad design, and allow changes that shouldn't be agreed.
I thought devs are supposed to be geniuses. Subnet splitting makes no sense. Splitting it in two just divides compute in half giving you two half as large blockspace subnets, i wouldn't need to test it to realize that its logic. Now I know I can dev a top 50 Project, thanks for making me realize my potential. The 1 canister limit for subnets is what made it so obvious cuz basically they are saying each subnet only needs one canister, if that changes I'll be shocked, but hey sharding surprised me so anything is possible. Maybe theres a way to implement it into ICP. With A.I. capabilities, infinite scalability has to be the ultimate goal.😊
I thought the same thing. How is it infinitely scalable but having issues scaling? AI will have a tough time running without resources. It sounds like some nodes were offline, so hopefully, they can add resources soon, and ICP can flourish
From what I understood of the forum post the single canister per user can be ok but Yral's canisters have a heartbeat so every call the subnet is loaded with messages causing congestion. Hopefully it's a terrible design decision on their part but it does highlight the current issues with how ICP handles this. Perhaps if they have a single canister to do this it would suffice instead of each user potentially issuing their own message to the queue. I would also note that this is a good stress test for ICP and I'd rather see these issues pop up now than during the peak bullrun
I am almost done with my application, multi canister, yes, they act as microservices, I am not creating dynamic canister per user, that is really silly even if it is allowed by the system.
Yeah. I'm also wondering what, if anything, Dfinity put into the recent Juno SNS launch attempt. Hmm.. I wonder if there's a way to look that up. Yet another thing on the to do list.
Yea, same. And I know it isn't "the fix", but something needed to be done. Also, it sounds like the sort of thing that isn't hard to change again when appropriate.
@@georgedcguy7402 have $13k invested in Icp @ $8.30. Would be happy to take our $30-40k this year. Lots of money to made out there and much faster too. also hold hbar and aero but have doubled and will eventually will buy BTC for early retirement lol
Ok so these guys want to be the new internet for the entire world and if that’s their goal and if their system can’t handle one single meme called bob, these guys are dreaming shit up for real. Wow!
I think this is a failure of having a DAO run ICP. Sometimes there needs to be a leader who understands the technical issues and lead the community as fast as possible instead of having to get the committee together to hopefully understand the issues and respond with the appropriate solution.
Yeah that's one of the issues. Dfinity has ~20% of the voting power, and then a bunch of people just follow their neuron. I wonder how much voting power they really have.
@@Coinopoly Probably a lot, but for example I'm followin over 10 neurons. If 10 neurons from different parties agrees on something, it is most likely ok.
Very legitimate concerns that needs to be addressed, we need more nodes
Jerry did a good one on this today as well, I'm sure you saw, I appreciate your added insight, what we need is Dominic Williams to address this and let us know in laymans terms, why this is occurring and what are possible solutions if any?
💯
Agree
I know about it, but I didn't watch that 1 hour video and didn't watch his other recent 4 hour video, either.
@@Coinopoly the four hour one is basically an e-book, very good
Adding more nodes to subnets will do nothing, it will be only be worse because update call has to reach consensus in the subnet which means every update call runs on all nodes anyway, so subnet is as strong as the one node strong. We need to increase node power or make switching canisters between subnets automatically, so you can scale your app running on multiple subnets without even knowing. Or we need to make sharding in the subnet it self to not run everything on all nodes, but that may be same as running canisters on multiple subnets.
Other thing that it seems that what overflows subnet is many small update calls because of additional work needed for nodes to handle those update calls, so maybe we need to increase base update call price, so developers will do less of small calls but more of bigger calls.
Awesome. That's one of the reasons I made the video; to help get people talking about it. And yes, obviously I am a content creator as well, so having something to talk about that might get people to watch is also part of the equation.
PLEASE COMMENT ON MY COMMENT A COUPLE THREADS UP... 😊We need to have separate chains for Trading (Memes creating calls) and possibly separate chains for token creation and actual subnet projects cuz memes only need social engagement, but if they need subnets, then separate Bob from icp-tv (a project i will be building in the future, but in this current structure i will have to expand my business and deploy on utopia which would suck as it should definitely be a community thing.)😢
They deployed a fix to one subnet yesterday and it seems to have fixed the issue. Im hoping they prioritise scaling a lot more now and forget about AI for a while
fixed the issue with Yral?
@@ASHdemoREEL yes, but only on one subnet. It looks like the fix went well and i can confirm the canisters me and my team deployed are not suffering the issue anymore. Over time the other subnets will get the same fix. However, DF still need to prioritise heart beats and the scheduler since the fix might not be the silver boot
Thankyou @@TheOmfg02 for your clarification, much appreciated. as an ICP developer, do you believe this is a simple solve for ICP? or could is this a dealbreaker? and could be the leak before the damn fully crumbles?
@@ASHdemoREEL there are a lot of options open to DF to improving the scalability issues. They just haven’t prioritised it for a long time. I understand they wanted to move on to different features and progress with their roadmap. This isn’t a deal breaker and im still confident they can improve things a lot. It’s not simple in terms of the time and complexity needed to create the solution but the solutions are possible with their current team quality, size and budget so i have good faith in them. It’s just a shame they didn’t do more testing to earlier to find this potential issue but again, it’s hard to create a testing environment that would expose the scaling issue
Thanks so much @@TheOmfg02 for your knowledgeable opinion. For current investors, is this fix solveable for the foreseeable current bullrun, or its a fix for the next 5-8 years?
None of this is even possible on any other chain
True.
Because none of other chains can do computation, same as wired home phones will never freeze compared to smart phones.
Thanks again for a great and informative video, I guess I was a little disappointed when I found out about the issues however I do think that in order for ICP to evolve into greatness, there has to be some growing pains...
Really good video, I wish Dom would address this instead of spending 100% of his time on AI. This is way more important.
@@jjwe2002 Dom is not concerned with retail and their faux pas. this is not a Dfinity problem, it's clear the devs are not versed on the Motoko language, yet. in the meantime let Dom focus on his target client base,.. enterprise, NGO's, military, and nation state contracts. gl
I love that question: why ain't nobody making proposals to add more subnets and nodes
because they are buying more ICP at this price before it will pump
This was brought up in 2021 and they said they were fixing it shortly, now a dev is saying that they shouldn’t build with a 1 canister per user model. Hate to be that guy but that’s the whole point of ICP. How many times has Dominic said each person will have their own AI running as a canister smart contract? I truly hope this gets resolved soon
A very similar thing happened to EOS. "Mining" projects hit it to extract resources and the chain was somewhat unusable until they came up with solutions. Growing pains and it certainly seems like these apps that did onboard a lot of users had a huge bill. Tough to come up with an appropriate funding model and a setup that can scale properly.
Thats reasonable critique , ty for showing this.
Developer may have had a poor design from the start. Canister is a Project (not a User). My worry NNS may not be aware of these shortfalls by bad design, and allow changes that shouldn't be agreed.
They need to add nodes, and improve hardware performance.
FreeDum Fighters' implementation of hierarchical ring signatures shows privacy protocol sophistication 🔐 🔄
They will work it out
Thanks for doing this video brother
I thought devs are supposed to be geniuses. Subnet splitting makes no sense. Splitting it in two just divides compute in half giving you two half as large blockspace subnets, i wouldn't need to test it to realize that its logic. Now I know I can dev a top 50 Project, thanks for making me realize my potential. The 1 canister limit for subnets is what made it so obvious cuz basically they are saying each subnet only needs one canister, if that changes I'll be shocked, but hey sharding surprised me so anything is possible. Maybe theres a way to implement it into ICP. With A.I. capabilities, infinite scalability has to be the ultimate goal.😊
Canister per user sounds really crazy from day1
Yeah, that sounded odd to me as well but I lack a lot of details and development skills since I'm not a blockchain developer or any kind of developer.
I thought the same thing. How is it infinitely scalable but having issues scaling? AI will have a tough time running without resources. It sounds like some nodes were offline, so hopefully, they can add resources soon, and ICP can flourish
Apparently it's due to heartbeat canisters. And no monitoring or automatic response of any kind. Which.. well, yea still an issue!
Very good content!
From what I understood of the forum post the single canister per user can be ok but Yral's canisters have a heartbeat so every call the subnet is loaded with messages causing congestion. Hopefully it's a terrible design decision on their part but it does highlight the current issues with how ICP handles this. Perhaps if they have a single canister to do this it would suffice instead of each user potentially issuing their own message to the queue.
I would also note that this is a good stress test for ICP and I'd rather see these issues pop up now than during the peak bullrun
Yea, I actually did see something about a heartbeat and totally forgot about it. I need to make a follow up regarding that.
Such a nice community!
I am almost done with my application, multi canister, yes, they act as microservices, I am not creating dynamic canister per user, that is really silly even if it is allowed by the system.
When it's done, let us know! I can always use something to talk about.
@@Coinopoly I am on testing period. Will inform soon. I will need sellers.
It will get worked out. Very informative, thank you. 🙏
Sponge V2 is catching a lot of attention-perfect time to buy before the next big move! 🌊💥
Sorry, what? Is this just the latest bot craze? I don't see a bunch of replies, so are you real?
nicely said
Superb narrative , wwll raised
Already fix in plan/place with vote middle of the week...
Hey we gave ETH years to get their network to what it is now. I am willing to give ICP at least that long since it is way more capable
Good video.
WHY FUD?
It's not exactly FUD, and I'll probably make a follow up video today or tomorrow.
@@Coinopoly Keep going with the truth. Crypto people are not used with that concept.
Thank you.
I wonder what has Difinity voted on this one...
Yeah. I'm also wondering what, if anything, Dfinity put into the recent Juno SNS launch attempt. Hmm.. I wonder if there's a way to look that up. Yet another thing on the to do list.
they voted YES
just voted yes
Yea, same. And I know it isn't "the fix", but something needed to be done. Also, it sounds like the sort of thing that isn't hard to change again when appropriate.
My friend this token don't move
🏎
more nodes will fix it
Solana guy 😂 or ether icp its so good that yeah looks unreal😊
If you mean I'm a $SOL or $ETH bull, you may want to browse the other videos on this channel.
growing pains..
Entire crypto need a refresh. Too many scams and too many projects that just BlaBlaBla
ICP❤
Give me $30 and I’ll sell out. This is going nowhere fast
impatient investors like you never make real money
@@georgedcguy7402 have $13k invested in Icp @ $8.30. Would be happy to take our $30-40k this year. Lots of money to made out there and much faster too. also hold hbar and aero but have doubled and will eventually will buy BTC for early retirement lol
weak hands, ICP is not for you.. ICP is a long term project read the road map.
No just sell so I can buy up your bag, you’re not meant to be an investor.
@@ImFAULKn take it for $30 anytime. You will never make money lol clown
Kaspa
Ok so these guys want to be the new internet for the entire world and if that’s their goal and if their system can’t handle one single meme called bob, these guys are dreaming shit up for real. Wow!
Pepe Unchained's meme contest winner gets 100k tokens, submitting my entry now. 🖼💸
Sorry, what? Is this just the latest bot craze? I don't see a bunch of replies, so are you real?
WTF
¿Que?
I think this is a failure of having a DAO run ICP. Sometimes there needs to be a leader who understands the technical issues and lead the community as fast as possible instead of having to get the committee together to hopefully understand the issues and respond with the appropriate solution.
That's why there is a following mechanism on dao. You can follow one of those leaders, who knows and understands the issues.
most people follow dfinity's neuron
Yeah that's one of the issues. Dfinity has ~20% of the voting power, and then a bunch of people just follow their neuron. I wonder how much voting power they really have.
@@Coinopoly Probably a lot, but for example I'm followin over 10 neurons. If 10 neurons from different parties agrees on something, it is most likely ok.
With the amount of $ I have invested, this hurts. Sometimes the truth does hurt. I hope they get this all worked out. Thanks for keeping it 💯!!
Apparently, it's not so bad at this point. Working on a video now but I'm having difficulty finding certain details. Owell, just another day.
Since day 1 I knew this can happen if someone really tested icp!
Now they can't flex their speed
Too many crushes. After a thousand canisters. Lol.