Steve Vinoski

It's not the scale, it's the reliability.

When I worked at Amazon.com we had a deeply-ingrained hatred for all of the SQL databases in our systems. Now, we knew perfectly well how to scale them through partitioning and other means. But making them highly available was another matter. Replication and failover give you basic reliability, but it's very limited and inflexible compared to a real distributed datastore with master-master replication, partition tolerance, consensus and/or eventual consistency, or other availability-oriented features.

I think everyone is missing the boat by evaluating datastores in terms of scalability. Even at Amazon's size, database scaling is only occasionally a problem. By the time your site is that big, hopefully you've split your software into independent services, each with its own (relatively) small database. And for the services that do need a distributed cache or persistence layer, there's now a huge body of literature and software that solves those problems.

Reliability is another matter. Every business can benefit from improved reliability, regardless of size. And reliability is harder. Scaling is basically an engineering problem, which can be solved with the right architecture and algorithms. Reliability is more like security: it's a weakest-link problem and requires constant vigilance in all areas, including engineering, testing, operations, and deployment. Adding redundant components improves reliability against hardware…

Hi,

Author of elli here.

It is very hard to create real and meaningful benchmarks. The "Hello world" benchmarks are very useful when writing a webserver and you are curious about where to optimize it, if necessary. Even though they are very superficial, they can also help you compare two webservers.

It is very easy to run the benchmarks on your own hardware. Get elli, then run "elli:start_link()" and hit "/hello?name=john" with apachebench or your tool of choice.

Elli is only useful if you want to write an Erlang application that exposes a HTTP API. If you want raw performance, Haskell has some servers which does 300k+ rps.

Here is a very good article from Steve Vinoski who is overall a very smart and experienced guy on the topic of benchmarking Erlang webservers: http://steve.vinoski.net/blog/2011/05/09/erlang-web-server-benchmarking/

Knut

Another nice example can be found in the last "Functional Web" column on IEEE Internet Computing by Steve Vinosky:
http://steve.vinoski.net/pdf/IC-Getting_Started_with_Google_App_Engine_and_Clojure.pdf

In the July/August 2010 column, Getting Started with Google App Engine and Clojure, guest columnist Aaron Bedra shows how to use Clojure, a relatively new but robust Lisp implementation on the Java Virtual Machine, create and deploy an application using the Compojure web framework on the Google App Engine platform.