Comments on: Scala actors for the enterprise: introducing the Akka framework http://blog.xebia.com/2009/10/22/scala-actors-for-the-enterprise-introducing-the-akka-framework/ Thu, 11 Mar 2010 16:23:01 +0100 http://wordpress.org/?v=2.8.4 hourly 1 By: Ulf Wiger http://blog.xebia.com/2009/10/22/scala-actors-for-the-enterprise-introducing-the-akka-framework/comment-page-1/#comment-93790 Ulf Wiger Mon, 11 Jan 2010 08:01:55 +0000 http://blog.xebia.com/?p=3264#comment-93790 FWIW, I used the phrase "let it crash philosophy" in reference to Erlang programming in a paper published March 2001 (http://www.erlang.se/publications/Ulf_Wiger.pdf). It may have been the first reference to the exact phrase in a paper on Erlang (I'm not sure), but it was cited as a common expression: "This, in combination with the use of pattern matching, leads to a style of programming that is often referred to as 'programming for the correct case' or 'the Let it Crash philosophy.' The latter is of course a bit provocative, but seems to serve the purpose of countering many programmers’ instinctive urge to save the life of their process at all cost." (p 14) Joe Armstrong's PhD thesis compares the concepts to "fail-fast" programming in the Tandem computers, and Ericsson had a form of "fail-fast" programming with fine-grained recovery, in the form of "small restarts", from the beginning (early 70s) in the AXE architecture. The AXE architecture and its in-service performance was the Gold Standard, and the CSLab experiments mainly tried to come up with a more modern and productive way of programming telecoms, without sacrificing the reliability. Anyway, good concepts deserve to spread. The idea of using process links for cascading exits came from Mike Williams back in the 80s, who was inspired by the "C wire" in the old mechanical phone switches - when you pulled the C wire, all resources allocated for that phone call were released. Adding the ability to trap exit signals allowed one process to supervise another. FWIW, I used the phrase “let it crash philosophy” in reference to Erlang programming in a paper published March 2001 (http://www.erlang.se/publications/Ulf_Wiger.pdf). It may have been the first reference to the exact phrase in a paper on Erlang (I’m not sure), but it was cited as a common expression:

“This, in combination with the use of pattern matching, leads to a style of
programming that is often referred to as ‘programming for the correct case’ or ‘the Let it
Crash philosophy.’ The latter is of course a bit provocative, but seems to serve the
purpose of countering many programmers’ instinctive urge to save the life of their
process at all cost.” (p 14)

Joe Armstrong’s PhD thesis compares the concepts to “fail-fast” programming in the Tandem computers, and Ericsson had a form of “fail-fast” programming with fine-grained recovery, in the form of “small restarts”, from the beginning (early 70s) in the AXE architecture. The AXE architecture and its in-service performance was the Gold Standard, and the CSLab experiments mainly tried to come up with a more modern and productive way of programming telecoms, without sacrificing the reliability.

Anyway, good concepts deserve to spread. The idea of using process links for cascading exits came from Mike Williams back in the 80s, who was inspired by the “C wire” in the old mechanical phone switches – when you pulled the C wire, all resources allocated for that phone call were released. Adding the ability to trap exit signals allowed one process to supervise another.

]]> By: Twitted by jboner http://blog.xebia.com/2009/10/22/scala-actors-for-the-enterprise-introducing-the-akka-framework/comment-page-1/#comment-93207 Twitted by jboner Wed, 25 Nov 2009 09:06:59 +0000 http://blog.xebia.com/?p=3264#comment-93207 [...] This post was Twitted by jboner [...] [...] This post was Twitted by jboner [...]

]]> By: Twitted by bubbl_scala http://blog.xebia.com/2009/10/22/scala-actors-for-the-enterprise-introducing-the-akka-framework/comment-page-1/#comment-93187 Twitted by bubbl_scala Mon, 23 Nov 2009 09:53:06 +0000 http://blog.xebia.com/?p=3264#comment-93187 [...] This post was Twitted by bubbl_scala [...] [...] This post was Twitted by bubbl_scala [...]

]]> By: Knowtu » links for 2009-10-26 http://blog.xebia.com/2009/10/22/scala-actors-for-the-enterprise-introducing-the-akka-framework/comment-page-1/#comment-92962 Knowtu » links for 2009-10-26 Tue, 27 Oct 2009 01:07:19 +0000 http://blog.xebia.com/?p=3264#comment-92962 [...] Enterprise scala actors: introducing the Akka framework | Xebia Blog (tags: software scala concurrency) [...] [...] Enterprise scala actors: introducing the Akka framework | Xebia Blog (tags: software scala concurrency) [...]

]]> By: Book Of JOSH http://blog.xebia.com/2009/10/22/scala-actors-for-the-enterprise-introducing-the-akka-framework/comment-page-1/#comment-92935 Book Of JOSH Sat, 24 Oct 2009 00:29:13 +0000 http://blog.xebia.com/?p=3264#comment-92935 The next generation enterprise Systems Of Service platform (the next JBoss if you want to go for it). - JVM based - Data feed from backend traditional J2EE+RDB or legacy applications -- the Systems of Record or Maintenance if you will - distributed multicore 2-3+ nodes => 32 - 96 or more cores, (soon to be) commodity box cluster - pluggable/dynamic/module substrate OSGI, e.g. Felix, Karaf - Components written in Scala -- w liberal use of Actors w OTP policies, some STM as well - Paxos based node coordination and eventual data consistency synchronization - The majority if not all data in memory, check pointed on SSD. - Cluster of N nodes survives failures and keeps on chugging. 2F + 1 = N - Composite App ready, as API is HTTP, REST, JSON/SimpleXML (JOSH) None of the commercial guys will go there. IBM, Oracle, the whole lot are way too invested in their J2EE + RDB offerings. None of their marketing guys will let them touch the golden goose until its too late. Meanwhile a decent solution that can receive receiving data from a mainframe or ERP, that can serves prices or item availability or coupons, or ... in realtime, millions of times a day on commodity boxes, no outage windows, no downtime and no invoice for annual license fees in excess of 25K per cpu core _has_ a market. Wide open ... the next BEA. Sheesh where the hell are all the VC dudes with a brain at least a tenth the size of his ahmmm wallet when you need them!?! Too many tweety social thingy clones still to chase I suppose. The next generation enterprise Systems Of Service platform (the next JBoss if you want to go for it).
– JVM based
– Data feed from backend traditional J2EE+RDB or legacy applications
— the Systems of Record or Maintenance if you will
– distributed multicore 2-3+ nodes => 32 – 96 or more cores, (soon to be) commodity box cluster
– pluggable/dynamic/module substrate OSGI, e.g. Felix, Karaf
– Components written in Scala
— w liberal use of Actors w OTP policies, some STM as well
– Paxos based node coordination and eventual data consistency synchronization
– The majority if not all data in memory, check pointed on SSD.
– Cluster of N nodes survives failures and keeps on chugging. 2F + 1 = N
– Composite App ready, as API is HTTP, REST, JSON/SimpleXML (JOSH)

None of the commercial guys will go there. IBM, Oracle, the whole lot are way too invested in their J2EE + RDB offerings. None of their marketing guys will let them touch the golden goose until its too late. Meanwhile a decent solution that can receive receiving data from a mainframe or ERP, that can serves prices or item availability or coupons, or … in realtime, millions of times a day on commodity boxes, no outage windows, no downtime and no invoice for annual license fees in excess of 25K per cpu core _has_ a market.

Wide open … the next BEA. Sheesh where the hell are all the VC dudes with a brain at least a tenth the size of his ahmmm wallet when you need them!?! Too many tweety social thingy clones still to chase I suppose.

]]> By: notfancy http://blog.xebia.com/2009/10/22/scala-actors-for-the-enterprise-introducing-the-akka-framework/comment-page-1/#comment-92932 notfancy Fri, 23 Oct 2009 17:19:26 +0000 http://blog.xebia.com/?p=3264#comment-92932 > Last week me and some of my colleagues had the pleasure of being on the > receiving end of an excellent training given by Jonas Boner. This is not professional, people (protip: it's Bonér). > Last week me and some of my colleagues had the pleasure of being on the
> receiving end of an excellent training given by Jonas Boner.

This is not professional, people (protip: it’s Bonér).

]]> By: Arjan Blokzijl http://blog.xebia.com/2009/10/22/scala-actors-for-the-enterprise-introducing-the-akka-framework/comment-page-1/#comment-92930 Arjan Blokzijl Fri, 23 Oct 2009 15:48:01 +0000 http://blog.xebia.com/?p=3264#comment-92930 Hi Albert, Akka is hosted on github, you can find it here: http://github.com/jboner/akka. You need to have git installed on your machine to checkout the source, for which you can use the following command: git clone git://github.com/jboner/akka.git. It builds with maven 2. In order to build it, you need to set the environment variable AKKA_HOME, and point it to the root of directory where you have checked out the sources. Cheers, Arjan Hi Albert,

Akka is hosted on github, you can find it here: http://github.com/jboner/akka. You need to have git installed on your machine to checkout the source, for which you can use the following command: git clone git://github.com/jboner/akka.git.
It builds with maven 2. In order to build it, you need to set the environment variable AKKA_HOME, and point it to the root of directory where you have checked out the sources.

Cheers,
Arjan

]]> By: Albert Strasheim http://blog.xebia.com/2009/10/22/scala-actors-for-the-enterprise-introducing-the-akka-framework/comment-page-1/#comment-92928 Albert Strasheim Fri, 23 Oct 2009 13:29:12 +0000 http://blog.xebia.com/?p=3264#comment-92928 Where and how did you download and/or build the version of Akka you used? Where and how did you download and/or build the version of Akka you used?

]]> By: J.F. Zarama http://blog.xebia.com/2009/10/22/scala-actors-for-the-enterprise-introducing-the-akka-framework/comment-page-1/#comment-92909 J.F. Zarama Thu, 22 Oct 2009 22:44:07 +0000 http://blog.xebia.com/?p=3264#comment-92909 Thanks for the article describing Boner's latest work; this material is a great reference to introduce concurrency and its implementation in Scala Actor's and now the proposed improvement as implemented by Akka. The minute I saw Boner's name I knew I had to read it and you have done a very good job explaining what, how and why of a subject that is difficult to understand and use for all of us but certainly more so for colleagues and personnel new to the subject; thanks again for a very good coverage of the topic. Thanks for the article describing Boner’s latest work; this material is a great reference to introduce concurrency and its implementation in Scala Actor’s and now the proposed improvement as implemented by Akka. The minute I saw Boner’s name I knew I had to read it and you have done a very good job explaining what, how and why of a subject that is difficult to understand and use for all of us but certainly more so for colleagues and personnel new to the subject; thanks again for a very good coverage of the topic.

]]> By: Raoul Duke http://blog.xebia.com/2009/10/22/scala-actors-for-the-enterprise-introducing-the-akka-framework/comment-page-1/#comment-92904 Raoul Duke Thu, 22 Oct 2009 19:04:49 +0000 http://blog.xebia.com/?p=3264#comment-92904 @TM Jonas Boner uh, by far he isn't the first and shouldn't be able to TM it i feel :-) http://www.google.com/search?q="let+it+crash"+software @TM Jonas Boner

uh, by far he isn’t the first and shouldn’t be able to TM it i feel :-)

http://www.google.com/search?q=“let+it+crash”+software

]]>