Hello! Today you may read here my interview to Christopher Meiklejohn, one of the speakers at the upcoming Erlang Factory in San Francisco. Christopher is working on Riak at Basho Technologies.
Erlang, Vector Clocks and Riak!
Paolo – Hello Chris! It’s great to have another Basho Erlanger here! Can you introduce yourself please?
Christopher - It’s great to be here! My name is Christopher Meiklejohn, and I’m currently a software engineer at Basho Technologies, Inc., where I work on our distributed datastore, Riak. In addition to that, I’m also a graduate student at Brown University, in Providence, RI, where I study distributed computing.
Paolo – Before joining Basho you were working in a different company (i.e., Swipely) where you dealt with Ruby code. Did you already know Erlang when you started at Basho? How would you describe the switch between these two languages?
Christopher - During the time I was at Swipely they had Riak deployed in production, which was what initially got me interested in Basho, Riak, in particular, Erlang. When I joined Basho, I knew very little Erlang and spent my first few weeks at the company learning it.
That said, I love Erlang as a language and as platform to build application on. I wouldn’t necessarily say that the change from Ruby to Erlang was anything that was unexpected, specifically because I already had functional programming experience using languages like Scheme and Haskell.
Paolo – Rubyists tend to be addicted to TDD. Were you able to maintain such a good practice also when coding Erlang?
Christopher - Well, I’ll start with a disclaimer. I was primarily responsible for the introduction of behavior driven development at Swipely for feature development, in addition to promoting pair programming within the development team.
That said, testing and verification of software is a very interesting topic to me.
While I believe that all software should be properly tested, I’ve never been particularly dogmatic about when in the cycle of development testing is performed: whether it’s done during development to guide the design of the software or whether it’s done afterwards to validate the authored components. I do, however, have one major exception to this rule: when attempting to reproduce a customer issue and validate a fix for the issue.
This is purely a pragmatic decision that’s come from working on large scale distributed systems: testing and verification of distributed systems is extremely hard given the number of cooperating components involved in completing a task.
At Basho, we take two major approaches to testing Riak: integration testing using a open source test harness we’ve developed that allow us to validate high level operations of the system, and QuickCheck for randomized testing of smaller pieces of functionality.
Paolo – At the upcoming Erlang Factory in San Francisco you will give the following talk: “Verified Vector Clocks: An Experience Report”. Can you introduce in a few words the arguments you will treat during the talk?
Christopher - My talk is going to look at an alternative way of asserting correct operation of software components, commonly known as formal verification.
The talk will specifically focus on modeling vector clocks for use in the Riak datastore using an interactive theorem prover called Coq. This allows us to assert certain mathematical properties about our implementation, and perform extraction of the component into Erlang codewhich we can directly use in our system.
Paolo – Who should be interested in following your talk and why?
Christopher - Given the topics involved, I’m planning on keeping the talk pretty high level and will touch a variety of topics: the theorem prover Coq, which implements a dependently-typed functional programming language, the basics of using Core Erlang, a de-sugared subset of the Erlang programming language, and how we put all of the pieces together.
Paolo – Lamport’s vector clocks are well known by people working in fields connected to distributed systems. Can you explain briefly what they are and in what fields they can be used?
Christopher – Vector clocks provide a model for reasoning about events in a distributed system. It’s a bit involved for this interview to get into the specifics about how they work and when they should be used, so I’ll refer to you two excellent articles written by Bryan Fink and Justin Sheehy of Basho.
“Why Vector Clocks Are Easy”
“Why Vector Clocks Are Hard”
Paolo – About the application vvclocks, are you planning to keep the development on? If so how can people contribute?
Christopher - At this point, the project mainly serves as a playground for exploring how we might begin to approach building verifiable software components in Erlang. What has been done so far is available on GitHub, it’s actively being worked on by myself as my time allows, and if you’re interested in helping to explore this further, feel free to reach out to me via e-mail or on Twitter.
Hello! Probably most of you already know the Erlang Factory Light will be held in San Francisco between 3 and 15 March . Among the 50 talks there will be the one held by Ferndando Benavides. Fernando is an Erlang developer from Argentina, currently working at Inaka as director of engineering. Let’s talk with him about Erlang, Inaka and his talk!
Let’s get started!
Paolo – Hello Fernando, thanks for making yourself available for the interview. Would you like to introduce yourself?
Fernando - Hi! Well… I’m an software developer from Argentina. I started programming when I was 10 using QBasic and “improving” gorillas.bas :P. From that point on I studied computing and programming my whole life until I graduated as a computer scientist in the Universidad de Buenos Aires, a couple of years ago. I started working when I was 18, coding in Visual Basic, then moved to .Net after several years. But thanks to my career path in the university I discovered functional programming (Haskell) and fell in love with it, so when I had a chance to actually work doing functional programming (this time in Erlang) I had no doubt I wanted to do so. That was 5 or 6 years ago. And I’ve been doing Erlang development ever since. About an year ago, I’ve been assigned as the Director of Engineering at Inaka. That basically means I’m now in charge of the architecture and design of all Inaka’s systems, whether they’re written in Erlang or not.
Paolo – You are currently working as Director of Engineering at Inaka: would you like to give some insight about Inaka?
Fernando - Well… you should check out our website: inaka.net
Inaka was found on September 2008 by Chad DePue, and I’m glad to say I was there. We were just 5 people working in a small apartment in Buenos Aires and we had just one project. Now we are a much much bigger team, with people working from different countries, in a big number of projects. But the idea behind our work was always the same: Our clients have beautiful startup ideas… and we do our best to turn them into real systems or applications that they can quickly launch and grow. Most of our products are iOS/Android applications with Erlang and/or Ruby servers on their backend. And most of the products we build have some important large-scale and/or high-concurrency requirements.
Paolo – How does Erlang fit in Inaka’s ecosystem? Why did you start using it?
Fernando - Erlang is our #1 language choice (at least since I’m Director of Engineering :P) for server developing.
We started using it since our very first system. That was a second screen system that provided additional content to display on iOS devices when a particular show was on air on TV. It also let users interact, by writing comments and “kinda” chatting. When designing the system, Chad detected the high concurrency and real-time demands that it had: during the hour the show was on air, lots of users would be connected and interacting and they devices had to be showing the right content in the right moment, not later. Erlang was the right choice for a language to write the servers with. BTW, making that system scale was not an easy task, even when written in Erlang, and that was what my last talk in an Erlang Factory was about
Paolo – At Inaka you have been using Erlang for many years. How would answer to this question? Do you confirm what Bob Ippolito wrote in his answer?
Fernando - Being a Haskell fan myself, I can tell you I nodded in approval on almost every paragraph he wrote, including the last one ;). So, yes… I confirm what he wrote.
As him, I’m also not aware of another platform that has the right combination of features, performance and maturity to make it possible to build the kind of systems that we build at Inaka. Maybe instead of binary matching syntax, I would mention that one of the main parts of most of our servers is a RESTful API that usually has a SSE component that keeps clients up to date with server pushes. How easy, clean and nice it is to write the code to implement that and make it scale in Erlang is hardly comparable to any other language that I know.
Paolo – In your talk at Erlang Factory Lite you will talk about: “Lucene Server: From Erlang to Java and Back Again”. Do you mind giving us some information about the talk?
Fernando - It will be a similar talk as the one I gave on the last ErlangDC. I’ll talk about a system that I’ve built and it’s now open-source, thanks to TigerText: Lucene Server. It’s an Erlang application that lets its users index, delete and query documents (i.e. proplists, in Erlang terms) using Lucene (which is written in Java). The key point here is that no Java knowledge is required for lucene_server users. All Java compilation, running, handling, stopping, etc. is managed internally by the lucene_server application. I’ll tell you how it’s works and how it’s implemented, and I’ll show you some lessons I learned along the way
Paolo – Who should attend your talk and why?
Fernando - My talk is mostly aimed at Erlang developers that once faced JInterface, got scared and then decided to implement their solution in a different way… I was there, too
Also, those devs that need a great indexing solution like Lucene as part of their systems. They will learn about an open-source application that may be exactly what they need.
Paolo – About Erlang and Java, I really liked your blog post on Inaka’s blog. Can you talk briefly about it?
Fernando - I wrote that post before lucene_server was open sourced and, in the same way that I couldn’t talk about the second screen system I described before but nevertheless I wrote a blog post about its story, I wrote this post (or rather this still unfinished series of posts) about lucene_server’s history.
You worked both with JVM and Erlang VM which are widely acknowledged among developers. Talking about performances and tuning, can you give some pros and cons for both?
I never feel comfortable talking about performances and tuning of VMs. I’m not sure I know enough about Erlang VM to talk about that and I’m positive I know nothing about JVM in this respect. Also, comparing the performance of two virtual machines for two completely different languages is tricky to say the least. It’s not like you can run the same code in both of them and see which one behaves better
Paolo – Last question: lately I have seen a lot of tweets complaining about the Erlang syntax. What do you think about this?
Paolo – BTW, I am really puzzled about your name: is it Brujo or Fernando?
Fernando - Well… According to my passport, I’m Fernando Benavides Rodriguez. According to anyone else, I’m just “Brujo” :). In other words, “Brujo” is my nickname, it means something close to “sorcerer” or “shaman” in spanish. The nickname was taken from my favorite character in a series of books I love (La Saga de Los Confines, by Liliana Bodoc) and it stuck with me forever. So, just call me Brujo
Hello! Today in this blog you can read my interview to Gianfranco Alongi. Gianfranco is a Senior Software Developer at Ericsson with a strong passion for Erlang, Extreme Programming and Agile Methodologies. I am sure you will enjoy his answers
Erlang, TDD, XP, agile….what else?
Paolo – Hi Gianfranco! It’s a pleasure to have you here today. Can you please introduce yourself to our readers?
Gianfranco - Hi, and thank you! I’m honored to be interesting enough to be mentioned in this blog!
My name is obviously Gianfranco, but my full name is Gianfranco Franco Alongi, the middle name is a relic from the laws regulating concatenation of first names.
I have always felt a deep satisfaction in programming and problem solving involving abstraction through rule sets. I get a feeling of emotional purification when crystalizing a concept, or even a crude feeling of something into a formalization that can be analyzed, executed and mechanized. I can only imagine this is what Vulcan ’enlightenment’ feels like, when you see a clear path through reality, a sense of very vivid focus and understanding.
I am a Senior Software Developer at Ericsson, doing Erlang and other things, such as lecturing about good Design, Clean Code, TDD and Software Craftsmanship. I get paid to do Erlang, but my cognitive breath mint equivalent is APL, Haskell or CLISP. I have lately got totally hooked on APL, and will travel to Florida the 19th to hold a workshop on TDD and present a Testing Library I developed for the APL community.
To me, programming is a hobby, and I get paid to do this hobby during the day, but do my hobby without pay during the rest of the day (don’t tell my managers: but I would probably do the same thing even if I was not paid ; ))
My mind is to intellectual challenges what a husky is to physical exercise, I need it every day, and a lot of it. If I go without it for too long, I get restless and depressed.
Paolo - Would you like to tell us something more about your personal experience with programming and especially with Erlang? How did you start using it?
Gianfranco - Ah, yes. I got into contact with programming through my father. He worked as an IBM mainframe developer for many years, and we always had lots of books and stuff at home. He had a whole shelf with RPG, Assembly and instruction sets for different architectures. However, none of this was interesting to me. I got going with Quick Basic in 7th grade (I must have been 12 at the time), and wrote some useless stuff like text adventures.
From there, it moved on quite steadily, and in high school (Gymnasiet in swedish) – I had a math teacher who handed me a (gasp!) Pirated Copy of a Redhat enterprise edition. This is where I got serious with C++ for a while. I was now 15 and got a lot of help by this man. I knew that this was what I wanted to do! This felt so right. The programming.
When I later moved to Gothenburg, and studied at Chalmers, I met John Hughes, a very influential man with a keen sense for Functional Programming.
During this time I did a lot of Haskell, it was the main weapon of choice, but in one of John’s courses, we just happened to glance at Erlang and this was enough to catch my interest. I got hooked. And since that day, I’ve been doing it. For some reason, I have never felt that C, C++, Java, or C# has the right ‘feel’.
My cognitive mind has chemistry issues when it comes to those languages, but Haskell, Erlang, CLISP and now APL has been like hanging out with a really good friend. It’s a silent understanding between two friends. It is beautiful.
Leaving Chalmers, I was contacted by Ulf Wiger, and I started working for Erlang Solutions.
Paolo - You had the opportunity of being an Erlang developer in two of the biggest Erlang companies out there: Ericsson and ESL. Can you tell us something more about these two experiences?
Gianfranco - Erlang Solutions is a company with a lot of young talents and a lot of opportunities. You can be sent to Mexico, work with people in the USA, and basically see and do anything, as long as you are willing and able. This was a fun and interesting time, moving to London, leaving Sweden and meeting a lot of new friends. As ESL is mainly a contractor company, you get to work on a lot of different projects, this suited me well, and I loved the high pace. It is an environment where the fast movers and strong fighters can grow quickly, accumulating a strong CV with various technologies.
Ericsson is different from this. It is a corporation, thus, by default, the pace is lower. Much lower. Ericsson does not attract the passionate developers as well as smaller companies like ESL, and there can be a lot of bureaucracy which prevents the fast movements I was used to. However, Ericsson, being a large corporation, has the money to support and lift the passionate developers on its strong shoulders, and it can be a very different experience based on different managers. All my managers have been great, and value skill and craftsmanship, giving me the opportunity to act with a lot of freedom – and the accompanying responsibility to go with that freedom.
At the end of the day, it is us, the developers who form the culture we live in, the wrong people can make gold turn to lead even in the best places. If we are willing to change ourselves, we can do anything.
Paolo - In your current position at Ericsson you are also teaching Extreme Programming values and principles. Even though I know Ericsson has a solid story on agile methodologies I have to ask: is it difficult sometimes to advocate agile principles in such a big company?
Gianfranco - Yes.
Like most large corporations, the ‘Agile Transformation’ is first hand an adoption of ceremonies and not really about the governing values and principles. What should be a common understanding and discussion on values and goals, becomes quickly an act of doing the right things, instead of reflecting on why we do things, and how we do them. My belief is that part to blame for this, is that Agile is being pushed down from the top, and not being something perceived necessary from the largest part of the organization.
I have yet to meet a developer who is positive about Agile. Few have even heard the term Agile Software Development, and even fewer know about XP. We have a large legacy of Telecom experts, and a dwindling few who are Software Developer Experts. As a result of this, few care how things are done, the majority cares for doing things. And, as you know – the price is paid in sweat, bugs, and blood.
The only way to go fast – is to go clean.
Paolo - In the company where I work, Kent Beck’s book on XP is a must-read book. Beck states that XP helps developers to deliver better software. What do you think about this sentence?
Gianfranco - I agree.
My only remark is that we must never get stuck in one particular school of thought and shall always explore, educate ourselves and invest in our own career as Software Craftsmen. It is not only a matter of professionalism, but also a matter of survival. By learning new practices and principles from others, if I can reduce my own cognitive load by using a certain technique, and therefore spend more time optimizing for maintainability, it is definitely worth the time and effort to learn.
We all leave a legacy behind us, daily. If I take an extra hour refactoring my code, after passing all the tests and removing
duplication, it may seem like a large cost, but I know that in my unit, with ~200 developers, if I can reduce the time they spend
reading to comprehend it, there is an economy of scale that makes a good case to optimize for maintainability.
To me, XP is a lot about being a good developer with great practices, delivering value and being a pleasant man to work with.
The ‘team’ is very important to me, and I believe the best place to work at, is a place where you feel that you go to work with friends, the kind of people you feel would have your back in a Zombie Apocalypse.
Paolo - Another great book by Beck is Test Driven Development By Example. When I read it I really appreciated the part about eliminating the fear using tests. I know you also appreciate this approach, so how would you encourage a developer to start adopting TDD?
Gianfranco - My best tip to learn anything, is to do it during dedicated practice time. Children are fearless learners, they just do, and fail, and do, and fail. Eventually the feedback mechanisms allow us to tune ourselves, and we start succeeding more than failing.
With dedicated practice time, there is absolutely no pressure to succeed, and so, we loose the fear. Fear is the mind killer that puts the mind into an accelerated mode of anger and self pity, instead of the relaxed state necessary to tune properly on the feedback.
- Do a Code Dojo together with team mates
- or just book a 2 hour slot alone, at home.
Start by reading about TDD, and then choose a small practice problem, maybe a Kata or any other problem you can think of.
The main point is that it is not the problem in itself that is important, it is the practice.
It does not matter that you do not finish anything, is it the journey, and the act of noticing what you notice, that is the important bit.
Once you feel that you have a grip on TDD, then you can apply it on production code.
Because if you have to get stuck, if you have to fail, do it in a safe place, don’t do it during work time.
Paolo - I appreciated very much your blog posts about TDD and Erlang. What are in your opinion the best tools an Erlang developer can use for testing his code? EUnit? Common Tests? Something else?
Gianfranco - I do not have a favorite, to me they are tools which apply well for their respective tasks. The only thing I am very careful about, is the feedback time the tool gives me.
I always use EUnit for Unit Testing.
Then I can use Common Test for Function Testing, but if the specific test case would take 20 minutes to get into place properly with CT, and I can do the same thing in 2 minutes using a bash script with netcat, then I go for the speed choice.
In my unit, we have a different tool which is based on top of CT, but with it’s own syntax. Mostly I try to avoid it, and the straw that broke the camels back, was when I could test a certain use-case by manual configuration and some bash’ing, in just under 10 minutes, and we had tried achieving the same thing using this tool, for 3 days.
Paolo - What do you think about continuous integration and continuous development? Do you recommend any specific framework or service for Erlang code?
Gianfranco - I think they are necessary practices that give a lot of value and make life easier. By exposing complexity early, and dealing with stress up front, we remove the pressure of doing integration and delivery later in the project.
No recommendation, sorry.
It is the doing of the work that exposes what needs to be done.
Paolo - Last question: emacs, vim or what else? :D
Gianfranco - Emacs, and vim when there is no emacs. I use this together with Tmux, and the no-mouse environment I use, enables a very high bandwith between my brain and the computer.
 Star Trek Reference
 such as TDD
 Yes, I did it, I stole the quote ; )
Hello! September 2013 has come and here I am once more to give you some information about another year of Erlang blogging
Would you like to know something more about what I posted in this blog during 2013? Are you sure? Well, just keep on reading!
I can’t recall every single post I wrote this year, so I have to ad lib during this summary.
Since I noticed that you readers appreciate my interviews to famous Erlangers I kept on writing this kind of posts. I think nobody will be offended if I say that two of my favourite interviews this year were the one to Kenji Rikitake and the one to Steve Vinoski. I obviously want to thank all the others Erlangers that accepted my request for an interview and I hope that you had the chance to read at least some of their inspiring ideas and thoughts!
As promised during 2012 I tried to transform this blog into a space where Erlangers can get useful information about Erlang projects, events and books. In this sense I had the opportunity to write my impressions about the book “Learn You Some Erlang” (thanks No Starch Press) and my thoughts about “Erlang by example with Cesarini and Thompson” (thanks to O’Reilly).
Some posts were as usually more technical, in this sense I enjoyed very much writing the one about TDD and kata, the one about the cost in terms of time you have in your supervisor when spawning multiple gen_servers. I also liked sharing with you the most interesting Erlang articles I read online lately in my recent post.
That all! As always I want to thank all you guys for your support and patience. Hope you will keep on following my blog during these last months of 2013 and during 2014! Have a nice Erlang year!
Even though I haven’t been writing in this blog very much during the last moths, I had the chance to read many new (and old) interesting blog posts, articles and tutorial about Erlang.
Some of the stuff I had the chance to read was really well done and inspiring, therefore I would like to share these contents with you. In this blog post of mine I am going to make a list of the 4 articles that I loved most, with a brief description for each of team. Let’s start then!
Interesting Erlang stuff you should take a look at!
Here is my brief list; as I did in one of my previous blog posts I would like to remember that the order of the links is totally random!
- Create and Deploy Your Erlang/Cowboy Application on Heroku in 30 Minutes: In this blog post, Roberto Aloi explains how to create and deploy easily a web application written using Erlang and Cowboy. I liked this article very much: Roberto is one of the most famous Erlangers out there and even though the application he presents is really easy it’s always a pleasure to read some of his code. Moreover the part related to deploy focuses on Heroku the well known cloud platform, showing how easy it is to deploy our Erlang application there.
- Continuous Integration for Erlang With Travis-CI: continuous integration (CI) is a powerful tool. While I was searching for a good CI service to be used with one of my Erlang projects I ended up reading this nice blog post by Ward Bekker. In his tutorial Ward explains briefly what are the main tool we have in Erlang to automate our tests and teaches how to connect a public GitHub repository to Travis-CI. I really liked this blog post, and if you are new to CI I suggest you to read as well this article by Martin Fowler: Continuous Integration.
- Meck and Eunit Best Practices: TDD and testing in general are one of the things I am trying to learn more in these months. Even though this post by David Reid is not very recent (2011) I suggest you to read it if you want to learn more about Meck, the most famous Erlang mocking library written by Adam Lindberg. Once more, if you are new to mocking and so on, I suggest you once more to read a great article by Martin Fowler: Mocks Aren’t Stubs.
- A Week with Elixir: That’s true, Elixir is not Erlang, but I think we all should read what the great Joe Armstrong thinks about this new interesting language written by José Valim. In his blog post Joe gives his impressions about Elixir and its syntax, providing code samples and really wonderful insights. If you want to know more about Elixir you can go to the official web page or consider buying this book by Dave Thomas.
And that’s all folks! I hope you will enjoy these stuff as much as I did!
Hello there! Today I want to introduce you my interview to John Koening. John is a PhD student at University of Minnesota in the fields of distributed and real time simulations. John is also working in the game studio he founded. Currently they are developing the game The Electric Adventures of Watt which has some Erlang in it.
Learning something more about Ymir
Paolo – Hello John and welcome to my blog! Can you please introduce yourself to our readers please?
John - Hi Paolo, thanks for having me. My name is John Koenig and I am a PhD student at the University of Minnesota (UMN) studying distributed, real-time simulation. I am going into the third year of my PhD program preparing for my written and oral defenses. I have been a regular Erlang-user for about 6 years.
Prior to, and inter-mixed with my time at UMN I worked at Cray Inc. Most recently I was contracted under Cray’s Chapel team where I worked on several language improvements in the area of portability. A majority of my time at Cray was spent as part of their Custom Engineering initiative. Together, we engineered unique super-computing platforms and software stacks for various customers.
In 2010, I founded a game studio, Called Shot LLC, with my good friends Gabriel Brockman and William Block. We are currently in the first round of funding for our flagship game title: The Electric Adventures of Watt.
Paolo – This is a common question I ask during my interviews: how did you start using Erlang? What are the features of Erlang that made you learn it?
John - I was first introduced to Erlang while pursuing my undergraduate degree at University of Wisconsin – Eau Claire (UWEC), I think it was around 2006. As part of a Programming Languages course we were tasked with picking a new language and implementing a solution to a sufficiently interesting problem which applied to the language’s domain. At the time, I was big into Plan 9 and distributed software in general so I chose Erlang and implemented a distributed prime number sieve.
Being more of an applied school, UWEC had me spending most of my time programming C and C++ and I remember being really impressed with how Erlang modeled processes and inter-process communication directly in the language. Once I got past Erlang’s syntax learning curve and my newness to functional programming, I found myself able to express distributed solutions very naturally in Erlang. After that, I was hooked. I picked up Joe’s book, Programming Erlang, and started keeping up with the Erlang community online.
I first started using Erlang professionally at Basho in early 2008 when I was brought on as a Reliability Engineer. I had thought, coming out of UWEC, that I knew Erlang fairly well, but I grew considerably during my six months at Basho. Justin and his development team are incredibly talented and being around that level of skill and enthusiasm was highly contagious. I remember that time fondly.
Paolo – Would you like to introduce and describe in a few lines what Ymir is? Where will Ymir be used?
John - Ymir is an open-source (GPL), cross-platform, distributed 3D game engine written in Erlang.
With the number of cores available to gamers on the rise, it is Ymir’s purpose to break games out of the traditional, single-core-dominate game-loop and, in doing so, achieve faster, larger simulations which grow in proportion to the number of available cores.
Paolo – Why did you decide to use Erlang for Ymir? Was there any other candidate language at the beginning of the project?
John - Ymir grew out of a desire to create a multi-player RPG that got away from the traditional client/server model. Myself and a few friends enjoyed online RPGs but didn’t enjoy the MMORPG scene. We were interested the approach of Neverwinter Nights 2, however, which featured smaller worlds developed and hosted by members of the community. Hosting of these worlds could get terribly expensive, as the worlds were hosted on a single server and, as their player base grew, admins of these worlds would be required to either co-locate their server or pay for expensive home internet access with sufficient upload speed. I set out to change this, wanting instead to see a game capable of simulating a world in a more peer-to-peer fashion. Namely, a game engine capable of utilizing the additional computational power and bandwidth present as players login to enjoy the simulated world.
I didn’t consider anything other than Erlang for this task. Along with OTP, Erlang is still the best language for distributed development as it allows me to focus more on the high-level challenges of distributed real-time simulation and less on the gritty details of implementing my own task-queues, inter-process communication, etc. This choice was further cemented when I proved that communication to port drivers, with minimal trickery, was sufficiently fast to support online rendering.
Paolo – Reading your paper I spotted many words often used among Erlang developers: scalable, soft-thread, message passing and minimal amount of synchronization. Would you like to discuss the meaning of each term with respect to Erlang and Ymir?
John - Game engines are traditionally frame-centric. Their primary goal is to compute and render frames as quickly as possible. Two aspects make this approach difficult when scaling over multiple cores: first, computing a frame is recursively dependent on the frames which came before it and, second, traditional spatial data structures used in collision detection require all game entities to be synchronized in order to function.
Ymir takes an object-centric approach and aggregates frames as quickly as possible. Game objects (entities) are represented as Erlang processes (soft-threads) and each entity is responsible for simulating itself locally. Discrete events (e.g. collisions, user-input) are modeled as messages which occur at a specific point in simulation time. As an optimization we allow entities to exist at various points in simulation time and to resolve events in the recent past by application of timewarp. Entities proceed through their local simulations, streaming updates to their physical state to relevant renderers. To enforce fairness, a sense of global time is defined as the minimum of all entity simulation times. This introduces a small amount of global synchronization as entities “vote”on the value of this global time through various shared ETS-based counters.
To break away from traditional spatial data-structures, Ymir applies map/reduce to spatial reasoning in order to achieve scalable collision detection. When simulating forward in time, entities volumetrically hash their physical extents against a fixed cube to various buckets (also soft-threads) and aggregate contacts which result from writing their latest physical states into each selected bucket. The act of mapping to spatial buckets is analogous to selecting nearest-neighbors (broadphase) and the buckets themselves compute points of contact for each pair of entities overlapping within its given volume of simulation space. In short, Ymir serializes only those objects which are sufficiently close together while permitting objects sufficiently separated in space to simulate unimpeded.
Using map/reduce in this fashion allows Ymir to scale-out over many cores very well. Currently, we are able to realize ~11x speedup in overall simulation time on 16 cores and sustained frame rates of ~500 fps. Ymir’s performance is dependent on many factors, however, chief among these is the degrees of freedom between entities. As entities are serialized based on spatial proximity, scenes where all entities exist in persistent contact are currently unable to obtain such lofty speedups. I am currently expanding our methods to better model persistent contact which will help Ymir obtain better speedups in these scenarios. Furthermore, using map/reduce to compute contacts works well locally or over low latency networks but as we scale up to many machines connected with higher latency other approaches will be needed. I am currently investigating network overlays between entities which capitalize on spatial assumptions present in game simulations.
Paolo – Do you have any partial results about Ymir our readers can take a look at? What kind of tests do you do on Ymir?
John - We are actively maintaining performance results on Ymir’s indiedb page, and as time permits I will be documenting Ymir more completely on our development blog.
This video showcases the three testing scenarios we used to gather our preliminary results. All three scenes are rendered offline using Ymir’s built-in support for Mitsuba. Parallel is rather boring to watch, but provides a best case performance. Cylinders features a stack of spheres falling onto an static array of cylinders and is more representative of the types of rigid body interactions one might see in an interactive game. Last, is Bounce in which spheres move randomly within fixed scene boundaries. Bounce is currently being used to measure Ymir’s performance as it relates to scene density.
Paolo – Is there any way our readers can contribute to the development of Ymir? Is there any fund-raising? Can other developers join the project?
John - Glad you asked, yes! We are currently on indiegogo seeking funding for The Electric Adventures of Watt which will be powered by Ymir. In supporting The Electric Adventures of Watt, contributors will be directly helping us mature Ymir into its first public release.
We will be advertising the public repositories for Ymir concurrently with its first official release. In the meantime, if developers are interested in working on Ymir, please, don’t hesitate to get in touch with me: john calledshot.org.
Paolo – You are a PhD student at the University of Minnesota and your studies are mainly focused on parallel and distributed real time simulation. Do you think Erlang could be widely used in these fields?
John - Without a doubt, that is what is what Erlang was designed for. There is even a precedent for using Erlang server-side for games: MuchDifferent and SMASH. I feel that as CPUs continue to have more cores and affordable CPU accelerators (i.e. parallela) become available, game developers will turn to solutions like Ymir to grow their games. Scalable, real-time simulation is not an easy undertaking and savvy developers will be looking for the right tools for the job.
As many Erlang enthusiasts know, there are at times substantial resistance to using languages that are outside of other developer’s comfort zones. This is especially true for academia. I can’t even count the number of times I have had to defend Erlang to my lab mates at UMN. That said, we are not expecting that all developers wishing to use Ymir will embrace Erlang, and we have on our roadmap to develop front-ends for languages most game developers will find familiar: C/C++/Lua.
Paolo – Did you find any help in the Erlang community? Did any Erlang developer give you feedback or support online during your development?
John - Several times I got frustrated with the performance of Mnesia and ETS for Ymir’s collision detection and I turned to the Erlang IRC channel for support and guidance. The Erlang community has been nothing but insightful and supportive every time I have turned to them. Although for the life of me I cannot remember that handles of those who offered help, I owe the Erlang community thanks.
We also owe special thanks to you, Paolo, for featuring Ymir on your blog, Peer Stritzinger for helping us reach out to the Erlang community, and to your readers.