April 30, 2010 at 13:02HighLoad ++ discussion, new architectures
The purpose of the conferences that we organize, including the conference of developers of highly loaded systems HighLoad ++, is the professional growth of each participant in the event.
Therefore, we are constantly trying to get better. And the main direction of the search for quality improvement is, of course, the conference program, its substantive part. The ideal event, in our opinion, is such that the highest-level specialists would come to learn something new and learn at it. So that Sysoev , Rybak , Gabrielyan , Zolotukhin, Segalovich, Anatoliks, Shetukhin after the event said: “Yes, it was useful!”. That will be cool;)
That is why we will try to make HighLoad ++ new this year. We will process all information from all previous conferences and publish it. We will decode a part, maybe we will make a book. And we will not repeat this year - we will focus on the new.
At the suggestion of Alexander Krizhanovsky and Andrey Pantyukhin, we will try to invite Linux / FreeBSD authors with stories about what work is being done in the area of operating systems support a large flow of requests, about internal algorithms (processing network interruptions, I / O scheduling, file systems, process scheduling, management memory, synchronization mechanisms in the kernel).
Ivan Zolotukhin received a proposal to refuse all reports in the form “How to correctly configure MySQL” and invitereal authors of real working systems . As Asko Oia showed how Skype users are registered at HighLoad / 2008, we need similar speakers who work in real large projects. It is planned to do a lot of work at the invitation of Western speakers. At the same time, we will try to cut off marketing attempts, yes, the English marketing report is still fascinating - but this is not enough.
Oleg Bunin threw in a new topic for discussion of HighLoad ++ / 2010 - “ new architectures ”. Here opinions differed:
Carte blanche at the conference will receive original research in almost any field, for example, research by the anti-DDOS team from Moscow State University: BGPv4 issues: peer authorization, some consequences of the unsuccessful use of flap dampening. TCP / IP Issues: Incorrectly Closed Connections and Their Consequences, What Vendors (Redhat / MS / Cisco) Write on This Subject and How to Better Actually. Illustrated tandem attack on TCP / IP configured "by Redhat". Some interesting aspects of the TCP SACK mechanism and related problems.
Therefore, we are constantly trying to get better. And the main direction of the search for quality improvement is, of course, the conference program, its substantive part. The ideal event, in our opinion, is such that the highest-level specialists would come to learn something new and learn at it. So that Sysoev , Rybak , Gabrielyan , Zolotukhin, Segalovich, Anatoliks, Shetukhin after the event said: “Yes, it was useful!”. That will be cool;)
How to do it?
That is why we will try to make HighLoad ++ new this year. We will process all information from all previous conferences and publish it. We will decode a part, maybe we will make a book. And we will not repeat this year - we will focus on the new.
At the suggestion of Alexander Krizhanovsky and Andrey Pantyukhin, we will try to invite Linux / FreeBSD authors with stories about what work is being done in the area of operating systems support a large flow of requests, about internal algorithms (processing network interruptions, I / O scheduling, file systems, process scheduling, management memory, synchronization mechanisms in the kernel).
Ivan Zolotukhin received a proposal to refuse all reports in the form “How to correctly configure MySQL” and invitereal authors of real working systems . As Asko Oia showed how Skype users are registered at HighLoad / 2008, we need similar speakers who work in real large projects. It is planned to do a lot of work at the invitation of Western speakers. At the same time, we will try to cut off marketing attempts, yes, the English marketing report is still fascinating - but this is not enough.
Oleg Bunin threw in a new topic for discussion of HighLoad ++ / 2010 - “ new architectures ”. Here opinions differed:
- This does not need to be done, as the people do not know the basics (Ilya Kosmodemyansky, Alexey Rybak );
- There are no new architectures (Alexei Kharlamov, Stanislav Vyazovetskov);
- There are new architectures and there is something to study: starting from the concept of cloud computing (Andrey Pantyukhin, Kirill Korinsky).
Carte blanche at the conference will receive original research in almost any field, for example, research by the anti-DDOS team from Moscow State University: BGPv4 issues: peer authorization, some consequences of the unsuccessful use of flap dampening. TCP / IP Issues: Incorrectly Closed Connections and Their Consequences, What Vendors (Redhat / MS / Cisco) Write on This Subject and How to Better Actually. Illustrated tandem attack on TCP / IP configured "by Redhat". Some interesting aspects of the TCP SACK mechanism and related problems.