Wir verwenden Cookies, um unsere Website optimal für Sie zu gestalten. Mit dem Bestätigen des Buttons "Akzeptieren" stimmen Sie der Verwendung von Cookies zu. Durch Anklicken der untenstehenden Checkboxen können Sie auswählen, welche Cookie-Kategorien Sie zulassen möchten. Durch Klicken auf die Schaltfläche "Mehr anzeigen" erhalten Sie eine Beschreibung jeder Kategorie. Weitere Informationen finden Sie in unserer Datenschutzerklärung.
Notwendige Cookies helfen, eine Website nutzbar zu machen, indem sie grundlegende Funktionen wie die Seitennavigation und den Zugang zu sicheren Bereichen der Website ermöglichen. Ohne diese Cookies kann die Website nicht ordnungsgemäß funktionieren.
Präferenz-Cookies ermöglichen es einer Website, sich Informationen zu merken, die das Verhalten oder Aussehen der Website verändern, wie z.B. Ihre bevorzugte Sprache oder die Region, in der Sie sich befinden.
2nd Teil of the Übereinstimmungsalgorithmen Also for today: A new kind of algorithm We will then next year, in January in the 2nd Termin the PAXIS and the RAFT and the question is first of all: What is the problem? By an Übereinstimmungsalgorithmus it is about the process of a value to a value of a value This value is very common This value of a value can be a question from a client. And then the next question is the question of this client and the next question is the question of the other client. Just such things. So that it is clear in the system in global system that this is the next value. It can also be a value of the decision-making process. Also that is relatively clear. The Paxos is a more than a decision-making process where you can easily see different values. The raft is more than the client-antrags. and they will be locked up the line. But the problem is basically the same. Now it would be a way of feeling with Fehler freedom, but we don't have a lot of fear freedom, because we have a communication medium where we have two types of mistakes. The first option is the easier to handle. The information is simply verloren. Then the customer knows the information that the information came not at. The worst thing we have to do for these algorithms, because we don't have to do it, because otherwise we don't have to do it, the information is corrupt. How can we solve it outside of some algorithms? some cryptographic methods, some signatures or something else. you can see if the news is corrupted or not. So that you have a public/private key infrastructure over the server or anything. The second problem is that one is the news. The second problem is the processes. There is also a simple problem when the process stops. We can't do it anymore. We can't do it anymore. We can't do it anymore. We're going to go out in a timeout, we don't get the answer. We're done. If the process works, but it works, but it works, but it works, it works, incorrect news, we are in the first case, in the extreme case, even cooperates with other failure processes. If our system is hacked, it is the worst, the Byzantine mistakes, and the ones are not in the grip. We will then also have the same algorithm for both algorithms. So, how can we deal with un-for-laces communication? Well, what we do typically do, we learn how to send a Quittance, that means I have a message sent to the Puppance and wait until I get the Quittance from the Empower and then I know my message is coming. If you have a business logic algorithm, that is a way to determine whether the news really came up and because of which he will do something to do, then he can work with these quitts. But the problem is, you can see yourself, if he doesn't get any Quittance, send it again, and so long will be repeated and so on. And the main thing is, here in the example, PI to PJ send, both want a Success function to the function, if they vote. And this is now simple, but actually, when you are in unbearable communication, where you are on this Quittance, is not to solve this problem. Um... Allgemein, in a separate environment, it is not possible for the P1 and PI and PJ, the current state of the moment to be in the way. And only to see how you can see this, how can you see this, is now the following idea. For the proof that I have these two processes, PI and I have PJ. Let me see if I'm in the way. So, PI sendet a "Nachricht" an PJ. PI will know if the message is coming. And so send PI a quittance to PI. PI receives the quittance and says: "Okay, my message is coming." is at PJ's time. The two of them are going to success, also this function, only then to do it when they both agree with the information, the message is coming. So that means PI knows now that my message is at PJ's time, but PJ knows if PI knows that the message is coming to him. That means PI should just simply send a acknowledge message and say: "PJ, your confirmation is coming to me, I know that you have received my message." Pj becomes now the best way and we know that Pj has the best way Um... PI knows that I got the message that Weiss aber PI jetzt... that I know that he got the message. And to make sure that PJ should also be confirmed, and that a acknowledge for the confirmation that PJ the news and the confirmation and the confirmation. So I get a knot in the mouth. They already know what it goes on. Um... It goes so, they send constantly messages back and forth. And the question is: Can it be a last message to that point? And at that point, are all satisfied. We could now say, okay, now it's a last one, M'. And now they know both of them, the each other has the news and they know that I have the news and so on. So the both of them are now over a. We're going to take a look at... That would be a minimal sequence. That is the first message, when both are satisfied. If that is the first message, when both are satisfied, then it is so that PI doesn't know if the message is at PJ. Because otherwise PJ So that means that you could leave this message again. And So that the original sequence was not minimal. So, to show you how to prove something like this. Well, yes, we have to talk about the Byzantine Fehlerm, that you heard that. Also, man spricht von Byzantine Fehlermermen, und das geht darauf zurück auf die Geschichte, was der Byzantinischen Generäle, da gibt es quasi eine Menge solche "Divisionen" heißt es hier, da ist immer ein General und in der Mitte ist dann eben das gallische Dorf oder was auch immer, was angegriffen werden soll. Und es ist so, dass die Generäle quasi sich übereinstimmen müssen, wann angegriffen wird, zu welchem Zeitpunkt, also morgen früh im Morgengrauen oder was weiß ich. Und deshalb gibt es jetzt zwischen diesen Divisionen gibt es dann irgendwelche Botschafter, das heißt, das sind Nachrichten, die können auch kreuz und quer laufen und so weiter. Ich deute das nur mal bei den Nachrichten. I'm going to turn the power back on the back. I'm going to turn the power back on the back. And that's the story behind it, where it came from.
