The founders of the theory of distributed systems in the arms of hydra

    This is Leslie Lampport - the author of the fundamental works in distributed computing, and you can also know him by the letters La in the word La TeX - “Lamport TeX”. This is the first time he introduced the concept of consistent consistency back in 1979 , and his article “How to Make a Multiprocessor Computer That Correctly Executes Multiprocess Programs” won the Dijkstra Prize (more precisely, in 2000 the prize was called in the old way: “PODC Influential Paper Award "). There is an article on Wikipedia about him where you can get some more interesting links. If you are enthusiastic about solving problems on happens-before or the problems of the Byzantine generals (BFT), then you should understand that Lamport is behind all this.

    He will also be coming to our new conference on distributed computing - Hydra, which will be held July 11-12 in St. Petersburg. Let's see what kind of beast this is.

    Hydra 2019

    Topics like multithreading are some of the most popular at our conferences, they have always been like that. It was just deserted in this room, but here a man appears on the stage talking about the memory model, happens-before or multi-threaded garbage collection and - boom! - already under a thousand people take up all available space to sit down and listen carefully. What is the essence of this success? Maybe in the fact that we all have some kind of hardware on our hands that can organize distributed computing? Or is it that we subconsciously understand our inability to load it at its true worth? There is a real story of one St. Petersburg quantum (that is, a financial quantitative analyst and developer), in whose hands was a computing cluster, the full power of which only he alone can use. What would you do

    Due to this popularity, the topic of performance and efficient computing tends to spread throughout the conference program. How many of the two days of presentations can be made about performance - a third, two-thirds? In some places, there are artificial restrictions that limit this growth: in addition to performance, there must still be room for new web frameworks, for some kind of devoop or architectural astronautics. No performance, you won’t eat us all!

    Or you can go the opposite way, give up and honestly make a conference that will be entirely about distributed computing and only about them. And here she is, Hydra.

    Let's honestly admit that today all calculations are somehow distributed. Whether it be a multi-core machine, a computing cluster, or a large-scale distributed service, everywhere there are many processes that simultaneously perform independent calculations, synchronizing with each other. Hydra will be dedicated to how this is arranged in theory and works in practice.

    Conference program

    The program is currently in the process of formation. It should include reports from the founders of theories of distributed systems and the engineers working with them on the prod.

    For example, it is now known about the participation of Leslie Lamport from Microsoft Research and Maurice Herlihy from Brown University.

    Maurice Herlichi is a very famous and respected professor of Computer Science, about him there is also a Wikipedia page where you can go over the links and works. There you can notice as many as two Dijkstra awards, the first for work on Wait-Free Synchronization , and the second, more recent, Transactional Memory: Architectural Support for Lock-Free Data Structures . By the way, links do not even lead to SciHub, but to Brown University and Virginia Tech University, you can open and read.

    Maurice is about to hold a keynote called "Blockchains from a distributed computing perspective." If interested, you can take a look at the recording of Maurice's report from St. Petersburg JUG. Evaluate how clearly and clearly he conveys the topic.

    The second keynote, titled “Dual Data Structures”, will be read by Michael Scott of the University of Rochester. And guess what - he also has his own Wikipedia page . In Wisconsin, he is known for his work as a dean at Wisconsin-Madison University, and in the world he is the person who, along with Doug Lea, developed those non-blocking algorithms and synchronous queues running Java libraries. He received his Dijkstra Prize three years after Herliha, for his work “Algorithms for scalable synchronization on shared-memory multiprocessors” (as it should be, it is openly in the online library of the University of Rochester).

    There is still a lot of time until mid-July. We will talk about the rest of the speakers and their topics as the program is clarified and approaches July.

    In general, the question arises - why do we do Hydra in the summer? After all, this is a dead season, holidays. The problem is that among the speakers there are university teachers, and any other time is taken from them. We simply could not choose other dates.

    Discussion areas

    At other conferences, it so happens that the speaker read what is needed and immediately left. The participants do not even have time to look for him - after all, without a gap, the next report begins. This is very painful, especially if such important people as Lamport, Herlichi and Scott are present, and you generally go to the conference only to meet with them and discuss something.

    We have solved this problem. Immediately after his report, the speaker goes to a special discussion area equipped with at least a whiteboard with a marker, and you have a fairly large margin of time. Formally, the speaker promises to be there at least the entire time between reports. In reality, these discussion zones can stretch for hours on end (depending on the speaker’s desire and endurance).

    As for Lamport, if I understood correctly, he wants to convince as many people as possible that TLA + is a good thing. ( Wikipedia article on TLA + ). Perhaps this will be a good chance for engineers to learn something new and useful. Leslie offers this option - anyone who is interested can see his past lectures and come with questions. That is, instead of keynote, there may be a specialized Q&A session, and then a dicussion zone. I googled a bit and found an excellent TLA + course (officially duplicated by a YouTube playlist ) and an hour-long lecture “Thinking Above the Code” with Microsoft Faculty Summit.

    If you perceived all these people as cast names in granite from Wikipedia and on the covers of books - it's time to meet them live! Communicate and ask questions that the pages of scientific articles will not answer, but their authors will gladly contact.

    Call for papers

    It's no secret that many of those who are reading the article are not averse to telling something interesting enough. From an engineering point of view, from a scientific point of view - from any. Distributed computing is a very broad and deep topic, where there is room for everyone.

    If you want to perform next to Lamport, this is entirely possible. To become a speaker, you need to follow the link , carefully read everything there and do it according to the instructions.

    Be calm, as soon as you join the process, they will help you. The program committee has the capacity to help with the report itself, its essence and design. The coordinator will help to deal with organizational issues and so on.

    Particular attention to the picture with the dates. July is a rather distant date for the participant, and the speaker needs to start acting now.

    School SPTDC

    The conference will be held on the same site with the SPTDC school, so for everyone who buys a ticket to the school, conference tickets - with a 20% discount .

    Summer School on Practice and Theory of Distributed Computing (SPTDC) is a school that provides a wide range of courses on the practical and theoretical aspects of distributed systems, which are taught by recognized specialists in the relevant field.

    The school will be conducted in English, so this is how the list of topics covered looks like:

    • Concurrent data structures: correctness and efficiency;
    • Algorithms for non-volatile memory;
    • Distributed computability
    • Distributed machine learning;
    • State-machine replication and Paxos;
    • Byzantine fault-tolerance;
    • Algorithmic basics of blockchains.

    The following speakers will speak:

    • Leslie Lamport (Microsoft);
    • Maurice Herlihy (Brown University);
    • Michael Scott (University of Rochester);
    • Dan Alistarh (IST Austria);
    • Trevor Brown (University of Waterloo);
    • Eli Gafni (UCLA);
    • Danny Hendler (Ben Gurion University);
    • Achour Mostefaoui (University of Nantes).

    You can freely watch the playlist of previous school reports on YouTube:

    Next steps

    The conference program is still being formed. Follow the news on Habré or in social networks ( fb , vk , twitter ).

    If you really believe in the conference (or want to use the special starting price, as they say, “Early Bird”), you can go to the website and purchase tickets .

    Meet me at Hydra!

    Also popular now: