ezmobius

gentoo hands down.

One big advantage is the runtime efficiency. Merb processes use less then half the ram of a rails process and can serve 2-5 times the traffic. This means you save a lot of money scaling your app for traffic demands.

Merb also does not lock you in to a specific way of doing things. It's very easy to bend the framework to your will.

Rails is great for the 80/20 rule. Rails will take you the first 80% faster then any other framework. But rails will make it super difficult to get that last 20% and will cost you double the hosting fees(over merb) or more when you have to scale.

My Startup(engineyard.com) has just finishing building our phone system in adhearsion. This handles our global support line with tons of features:

Provides a complete IVR written in Ruby, with helpers for menus, multiple dialing, etc. Provides a complete user interface written using Ruby on Rails, for an easy to use configuration tool for both users and administrators. Supports PSTN numbers as extensions, providing a completely virtual PBX. Supports automatic creation of workgroups, including insertion into the dial plan, right from the Ruby GUI. Supports workgroup scheduling from the user interface, so you extension only rings when you tell it to. Supports dialing multiple phones and employees to answer an inbound call. Forwards voice mails to everyone in the workgroup Supports click to dial, call history, etc. from the GUI

The cool part is we're going to open source this thing in the next few weeks. It's probably the biggest adhearsion app out there yet and will serve as a great example for folks looking to do cool stuff with adhearsion.

There is a nice article on MVC<->MVC where you have your MVC server setup like rails or merb and you have an MVC in javascript as well. Good read: http://welcome.totheinter.net/tutorials/model-view-controlle...

merb

umm rubyforge.org?

XMPP pierces NAT for one reason. This means you can have XMPP clients behind many more firewalls then you can with amqp.

Also xmpp libraries are available in almost every language and platform and traditional MQ's do not have presence or federation(ie horizontal scalability).

It's not about using XMPP as a message queue per say. It's about using xmpp as a p2p messaging system for machine to machine communication as well as integration with chat.

These slides are totally offbase about static files with nginx and memcached. I've benched large nginx clusters storing static file page caches in memcached against just serving them from a GFS filesystem and giving linux lots of ram for buffer cache. Got about 8kreq/sec via nginx+memcached but got 23kreq/sec from nginx+filesystem with lots of buffer cache.

You're doing it wrong if you don't build your own framework at some point.

Blindly using rails or even merb or any other framework will only get you so far. To truly transcend, you must build your own or heavily alter an existing framework.

This is hacker news right? Go out and build your own framework already dammit!

what if your product is the company to sell?

We leached off of quality humans inc, which was a rails consulting shop some of our founders ran and also used savings and personal credit to get through the first 6 months or so after we all quit our day jobs ;) But lots of the planning stages, from Feb06 to August06 happened while I still had a full time day job somewhere else.

Well we were starting a new clustered, virtualized hosting environment so we had hard cash needs to get our first set of servers and equipment and data center space. So it's a little different then starting a software company without the same hard costs.

We basically built something like ec2 but with more advanced features like load balancing and clustered, posix compliant filesystems. And we shipped before ec2 came out or was even known about publicly. This kind of system costs $$ to do right so the 120k went mostly to hardware and data center costs to bootstrap us to the point where we could take customers.

We started planning in february of 2006 and shipped in september 2006 and it's been a wild ride since then.

ps - sure, no problem. Let me know if anyone has any other questions.

pps - yeah one of our founders started Parallax who made the BASIC stamp.

While this all may be a good plan it is not the only way to go. My startup, engineyard.com is only two years old. But we built something people wanted and started selling it to them in September of 2006 with only $120k seed money raised from friends and family. We were profitable by January 2007 and had 50% growth month over month for most of 2007.

We could have just left it at that but we had plans to become an open source company as well as a hosting company. Developing new software projects like Rubinius, Merb and Vertebra take time and money to pay top notch engineers. So we decided to take a gamble on going big and we took some VC money. The thing is, we already had a profitable business and the VC's came to us, we never sought them out.

So we were able to accelerate our strategy by taking the VC money and we got excellent terms since we already had a profitable business. And we took money from Benchmark that was less then other VC's were offering us because Benchmark's contacts and knowledge has been a huge strategic advantage for us.

We have been able to grow our 3 person startup to over 80 employees worldwide, and through our VC's contacts were able to secure a series B from Amazon, NEA and Benchmark.

We would never have been able to accelerate our growth and ideas as fast as we have if we never took money. By taking money we have cemented our place in the Ruby landscape as the 800 pound gorilla ;)

So yes maybe it is the safe play to not take money and try to sell a product that people want and thats great! But sometimes it really is worth taking big risks with big money working toward big goals.

This is not an either or scenario and both ways of doing things are equally viable

Someone gets it ;)

Amazon has stated that simpledb does not run on dynamo though so this is all moot.

We have a branch of Rubinius that is compiling ruby to rubinius bytecodes and then transforming that into LLVM bitcode for native compilation. It is still in the early stages but it already giving big speed improvements.

http://github.com/evanphx/rubinius/commits/cpp

I'd recommend github for source hosting. I've got multiple open source projects that I run or use that use github for scm and use lighthouseapp.com for ticketing. These are fairly large OSS projects with many many contributors and it works great. http://github.com/wycats/merb-core/tree/master and http://github.com/evanphx/rubinius/tree/master as well as a bunch of other smaller oss projects.

Both github and Lighthouse will give free accounts to open source projects.

now try this on for size. I'm sure you know how this works:

  > def a
  >   yield
  > end
  => nil
  > a { 42 }
  => 42

But did you know this?

  > def b
  >  Proc.new.call
  > end
  => nil
  > b { 42 }
  => 42

Proc.new without any arguments works the same as yield. tripped me out when I found this the other day.

that has to be a math mistake. With only 8916 tips, that dollar amount would require each tip to average $1132668.28 which doesn't seem to hold true...

Actually Vertebra is not tied to Ruby on Rails at all. Since the main communication is all XMPP based, any language that can speak XMPP that implements our protocol can become part of the XMPP cloud. Our first implementation of the agent framework is written in Ruby but I expect many more languages to be supported.We will be releasing at least python and erlang based agents along with our ruby agents. We will be releasing the whole thing open source once we have locked down the protocol and made it a XEP standard, so keep an eye on my blog if you are interested.

Also Vertebra is not just for server automation, it is a framework for distributed computation of many kinds as well as a real time application platform.

The discovery mechanism and Resource based dispatch will allow for applications to get a request, discover via a Vertebra call which backend is least loaded and offers the request processing you need and then dispatching said request to the proper backend. Also allowing for map/reduce and scatter/gather style programming.

I have many ideas of where to take this thing and I cannot wait to get it out in the open so I can see what other folks do with it.

get supermicro hardware. siliconmechanics.com is a pretty good provider of pre-assembled supermicro hardware. We use supermicro servers for all of our clusters at engineyard.com

gentoo on ~1800 virtual machines.

I like to write my slides and presentation the night before my speeches in the hotel at the conference. Helps keep it fresh. Say you have 50 minutes to fill... if you know your subject make about 30 slides and just be able to bullshit for 1-2 minutes per slide and you will do great!

You are playing right into this troll's hands. Don't bother responding to him, just delete his posts and ignore him.

I like lightspeed but the free version is crippled and only usable for smallish sites. The real version costs a lot of money but is pretty nice. But it not being open source makes me personally cringe ;)

Lightspeed is nice for smaller sites because the processes can go away when there is no traffic. But I'm not a fan of it much and it's not covered in the book.

Actually the book covers both apache and nginx equally so you should be straight. Although I must recommend that if you have a choice then you go with Nginx. Nginx is much more scalable and ues much fewer resources then apache does. The only reason to use apache with a Rails deployment is if you have other legacy sites to support that run on apache. If you are just deploying a Rails/Merb app then Nginx is far superior in every way.

You're more then welcome to stop by the Engine Yard offices if you wanted to. We will be having a hack night on tuesday ~8:30pm. We're located at 82 South Park St.

Gentoo all the way.

I apologize, I should have said that this is my startup. I've been a long time reader of new.yc but never felt the urge to post. But I am pretty excited about what we are doing and so I was compelled to respond.

My name is Ezra and my blog is http://brainspl.at if you have any interest in what I work on.

The most exciting thing about what we are doing wasn't really covered in this article. The real news is about our Rubinius(http://rubini.us) and Merb(http://merbivore.com) projects. We are trying to improve Ruby as a platform for everyone.

You also need to factor in redundant SAN, hardware load balancers, Master-> Slave mysql databases that do not use your slices resources. Clustered subversion, git and smtp, dns, none of which runs in your resource allocation. Databases are 4 core 4 gig ram for the master and for the slave. Offsite encrypted S3 backups every 4 hours. Worldwide team of expert rails support staff.