Topic: [XCR] Crypti | Dapps | Sidechains | Dapp Store | OPEN SOURCE | 100% own code | DPoS - page 60. (Read 804695 times)

Thanks a lot for your detailed explanantion !

it depends on the dapp.

If you have a heavy database driven dapp (e.g. chat service) they add very little, as only the master nodes will save the messages into the sidechain. 

If you have a service which is mostly viewing (e.g. news portal) then they add a huge scalability bonus.

Good design! Those "full nodes" provide "scalability" for hi-loaded dapps with million users 

Check here: http://blog.crypti.me/cryptis-dapp-platform-explained/

Quick explanation: There is the dapp owner (Founder). Several people (Crypti accounts) can connect to his account, making a multi-signature account. Those will be the master nodes, because they also have rights over the funds of the dapp and they take care of the database operations. (Consensus, similar to Crypti delegates) Full nodes just make the user interface of the dapp accessable to others (and to themselves).

When I say they have rights over the funds I mean this: e.g. there are 6 accounts, you need 3 OK to transfer the funds. Check out this blog post: http://blog.crypti.me/an-introduction-to-our-multi-signature-implementation/

When I say database operations, I mean they put the transactions into the blocks, forge new blocks, taking care of consensus in the sidechain. Working similar to DPOS. http://blog.crypti.me/introduction-to-dpos/


Regarding blog post, I would need to let the team check the above. Maybe I made a small mistake somewhere, but in general it's correct.

Passion_ltc what is the difference between "master nodes" and "full nodes" ?

Good post for blog by the way.

Hey sorry, yes I changed it a bit. Normally nothing should got pushed further back, maybe I made a mistake editing them or you saw a cached version. I will let Boris take a look over it and update the dates accordingly. I'm pretty sure it's not correct at every point.

I will tell you, when you can take a look again.


Edit: I see now, multi-sig is done already. But I pushed it back to be in one line with the sidechains and dapp VM as they will get released together. Avatars/Messages will be released once the store and everything is finished, bc they will actually get implemented as a dapp. About "Compute nodes integration", "Storage nodes integration" and "3rd party chain integration (BTC, etc...)" I just asked Boris. Let's wait for his opinion.

A dapp consists of several parts:

- storage of dapp files

Here we have two choices, one centralized (GitHub), but also a 3rd party decentralized method. The decentralized method can't be disclosed yet, but as you probably know there are several options on the market. At the beginning it will be one of them, and later we will add the others step-by-step. Maybe even an in-house solution will be coded as a dapp later by some developer.

- computation of the dapp

While a centralized application relies on a single server to perform it, we have 16 master nodes. This number will be raised in the future, but we will begin with 16 for now. If you are a dapp owner you can now look for 15 other independent persons, who can help you run your dapp.

- hosting of the dapp

Where is the difference between hosting and computation of a dapp? A normal server usually takes care of both, a master node at Crypti takes care of both as well. But an unlimited number of full nodes can install and start the dapp as well, so other users (lite client, mobile client) can access it though them.



Let's make an example, our example case will be a Bitcoin dice gambling game.

The team consists of 3 persons; a CEO (C), a technical guy (T) and a marketing guy (M).

Centralized (worst case):

The centralized service now has 1 server running for the website and C has access to the fundings.

- website hacked: All funds are stolen, the users will not get reimbursed.
- server down: The users don't have access to their funds.
- C runs away: The users lost all their money.


Centralized (good case):

The centralized service now has 3 servers running for the website and C, T and M have access to the fundings, within a multi-signature account in the cold wallet.

- website hacked: The hot wallet got emptied, users will maybe get reimbursed.
- servers down: The users don't have access to their funds.
- 2-of-3 admins run away: The users lost all their money.

You see, the good case is much better already. Now the advantages/disadvantages between centralized <-> decentralized services come into play. They already use a multi-signature account, but the servers can still go down.


Decentralized:

The decentralized service is running on 10 master nodes and 20 full nodes. C, T and M have 1 key each. But they also asked 4 delegates of the Crypti network, 1 Crypti foundation member and 1 escrow person to hold keys. This can of course also be done with a centralized service, but at Crypti there is automatically a verification of this possible.  Also it's inbuilt, resulting in a lot lower work for the site admins.

- website hacked: Well, the website can't get hacked. The only thing which can get hacked are the master node keys for Crypti and the decentralized hosting solution. Hot wallets might get emptied, but if the multi-signature implementation into the dapp got properly done, it's nearly impossible to do so.
- servers down: For this all 10 master nodes need to get DDOS'ed. This is of course possible, therefore the dapp owner could raise it to 16 master nodes. Later we will raise this number, so a dapp can get computed by for example 100 nodes. But the master nodes only take care of the database operations, for only viewing the site there are still many more full nodes. If you run your own full node to access the service, you will most certainly always have access.
- admins run away: Thanks to the multi-signature it's not possible. Again, this can be done in a centralized website as well. But at Crypti it's automated and checkable.

Along side these, the dapp is running on a sidechain. Here comes the regular blockchain technology advantages into play now. Everyone can check the liquidity of the service, can check the transactions on the sidechain and so on.

Crypti will also offer many more advantages, like
- an inbuilt token system (similar to the Nxt asset exchange, only for the dapps).
- APIs for Crypti and Bitcoin implementations.
- libraries and developer tools.
- a Dapp Store, making the dapps accessable from anywhere and from everyone.


I hope this explanation helps. I might have missed some things, even I can't imagine yet.

Damn. I have had this in mind because yesterday the roadmap were different. So you updated it few hours ago and then say like "I don't know why 50cent_rapper has these 45 days in mind". You have a sence of humor

Well new roadmap says DappStore will be in october  

Can someone explain to me what it is about cryptis dapps that is actually decentralized ? From what I gathered one person owns the app and it relies on that same person to run masternodes. Am i missing something ? Did i missunderstand that part ?

This is an honest question. I don't mean to hate. Just asking a question.

You will be able to use Cryptis or Bitcoin consensus. That means either 10s or 10m. But I think Boris mentioned once, that you can also use only every 10th Crypti block or so, making the block time 100 seconds effectively.

Regarding the second thing, we will when we release it. I think it's more important to market outside of BitcoinTalk, because we want to build the platform with many dapps creating value (=price) this way. We don't want to pump the price with announcements within the Bitcoin community.



Is there a quick way to do this with Ghost (blog software)?



I don't know why 50cent_rapper has these 45 days in mind. But we have a roadmap, check the OP. Here is the detailed one: https://docs.google.com/spreadsheets/d/1Zzsu0BikplztWB8xTU3PhMHbyHBrW8EX3lR-e6lS8LA/pubhtml

I probably need to explain it,

Multi-signature: The backend is done. The frontend will be finished this month. But, this is a feature which will probably be used by <1% of our users, therefore we will release it together with the dapps in v0.4.0. Else we would need to test everything again and the releases will be pushed further behind.

Dapps Virtual Machine: All dapps will run in a virtual machine (I hope you guys read the white paper. ), this functionality is done already. But this is only a part of the dapps, there are many more things which are done, but also many more things which need to be done.

Sidechains: Formerly we always used the term "Custom Blockchain", but the majority inside the cryptocurrency community call it sidechains. Therefore we changed the term for ourselves as well. They will act like decentralized databases for the dapps.

Avatars/Messages: These will be implemented as dapps, we don't want to pollute Crypti's main chain with them.

Also the sheet just talks about the crypti core/backend. The frontend is running seperately, sometimes it's finished much sooner, sometimes it takes 1-2 weeks longer. Also the public beta versions and test phases are not included in the roadmap, we will have to dramatically test the dapp functionalities.

Is there a road map? I haven't seen any particular timing.

I wouldn't mind if you enable anonymous comments under blog posts.

Google likes very much (in terms of SEO) unique content, regual updates and comments under this content.

A little more patience. Let's wait for 0.4.0 version with DappStore and custom chains - only 45 days to wait. And then start to market / advertise Crypti in media, not only cryptomedia, but in "external world" too. We need dapp studios.

I think angel investors will fund dapp studios / enterpreneurs, which will build decentralized dapps, for example, personal finance dapp, CRM system for business, encrypted chat. 

If you are Cryptillionaire, in the future you can fund those kind of dapp studios / enterpreneurs with XCR.

And bitcointalk - the majority of users there are intersted in converting electricity bills into BTC, like converting 150.00 $ electricity bill into 0.50BTC worth of an altcoin, which cost the same 150.00 $ by the way.

Max, your dapp overview is very good and still has me thinking.  

Is it possible to write the XCR code so master dapp nodes have an adjustable time (longer that 10 seconds) to perform side chain block addition?   Say a dapp author was OK with modifying the side chain every 30 seconds or once a minute.  This would probably overcome the latency problem that disqualified using home PCs / Raspberry Pis on slow home networks that were not hooked up to an internet backbone.   The ability to run master dapp nodes and side chains on $35 Raspberry Pi 2 quad-core computers at home instead of just at expensive server farms would be a big plus to Crypti.   Many more users would try to run Crypti dapps if it could be done on cheap computers running on home networks.  We had hoped to do this with the 10 second XCR main chain and it didn't work.  We should try to make this a possible option on dapp side chain master servers.  

A second thought - maybe we should do a whole new [ANN] announcement here on BTC for XCR dapps, to let people know that XCR has revolutionary new capabilities beyond the typical cryptocurrency.  

Please retweet the tweets!

https://twitter.com/CryptiTeam/status/619231180915900416
https://twitter.com/CryptiTeam/status/619231341494800385

Thanks for the details, Max.  This is a very good and very important overview.  I will need a little time to read your words and come up with my next round of questions ;-)

In the meantime, here is an article on brain wallet cracking worth reading, since that's what XCR uses:

http://www.wired.com/2015/07/brainflayer-password-cracker-steals-bitcoins-brain/

Or you can just skip the article and go straight to Awesome Password Generator set for 50+ characters (like I do):

http://code.google.com/p/awesome-password-generator/

The process is the following:

1.) Upload dapp

You upload the dapp to Github or the 3rd party decentralized storage solution I mentioned earlier.


2.) Registering dapp X from account A

When registering the dapp you have to specify some information, like name/category/source and so on. Account A can be a multi signature account with up to 15 other accounts. As the dapp X got registered within Cryptis main chain (for a fee in XCR) from account A it will be the "king of the master nodes". All multi signature accounts below account A, let's call them account A.1, A.2, ..., A.15, are the other master nodes of the dapp X.

Master nodes are independent of the main chain delegates. They can also be main chain delegates, but don't have to. You can simply imagine them as new delegates for the side chains. Taking care of the consensus there.

You can remove and add multi signature accounts, i.e. the master nodes are exchangeable at any time, of course only with enough confirmations from the other multi signature accounts.


3.) Install dapp X

You can now install the dapp, as it was registered on the main chain. The installation is only possible from the full clients. While the installation, the dapp source code will be downloaded from the source (Github or decentralized solution) and get copied into the Crypti directory, in a special folder for the dapp there.

You now see that the master nodes all have to be full nodes as well, and they all have to install the dapp. They also have to specify their passphrase in the dapp config, to be authorized as master nodes (only possible for multi signature accounts A.1, ...).

- The full clients are executing the dapp source code themselves, after the installation on their node of course.

- The lite clients are working via the API only. They need to pull the dapp informations from another node (both, master and full node possible). It's the same principle like pulling the regular Crypti data from full nodes today.

- The mobile clients are working the same like the lite clients.


4.) Starting / Stopping the dapp

From the full client you can now start and stop the dapp like a regular application. If it's running other users can access it, if it's allowed by the node owner.

The master nodes have to run the dapp all the time, same like delegates have to forge all the time. Because they also take care of the consensus of the side chain. The consensus is working very similar to the delegates on Cryptis main chain, i.e. the master nodes are creating blocks in the side chain. They also get the fees for the transactions within the side chain.


5.) Accessing the dapp

If the node is running the dapp, you can access it with opening the following address: http://:/dapps//

You can use your web browser, the Crypti Dapp Store or a electron/nw.js wrapper. The user interface can be similar to modern web apps, thanks to JavaScript. (If the dapp is well coded, it has the same user experience like Slack, Aether, Visual Studio Code, Crypti itself and many more..)


Conclusion

So optimally 16 master nodes are running the dapp, creating an untouchable base and providing the consensus (make new blocks). Additionally an unlimited number of full nodes are supporting the network of the dapp, for delivering a decentralized accessing option for the lite clients. To make a comparison again; the master nodes are the backbone of the whole system. The full nodes are acting like accessing servers, with the regular users accessing them with their lite and mobile clients. For now 16 master nodes are far more than enough, because they only have to take care of the consensus. It's not such a heavy task, at one point we might raise the number of possible master nodes.



I didn't understand all of your questions. I think you now have a much better image of the whole system and its functionality, if you still have questions you can probably much better describe them now. I will include the above in the status update of this week, I think it helps other users to understand our vision as well.

Crypti need only 1 super successful dapp in order to skyrocket to billion dollar marketcap.

But in order to find this 1 killer dapp a lot of other dapps must be roll out, games, chats, e-shops, crm...

Crypti needs to support Uber.

http://www.coindesk.com/uber-denies-plans-to-accept-bitcoin-payments/

Bicoin continues to flounder.

http://motherboard.vice.com/read/the-mystery-behind-the-biggest-bitcoin-transaction-ever-made?trk_source=homepage-lede

I have the time to do it later today. It will be a status update.