big erlang web solution

[Christian S]

On 6/22/06, bryan rasmussen <rasmussen.bryan@REDACTED> wrote:
> Hi, I've decided to work on a new web site project and I've decided to
> use Erlang for it.
>
> The reason for using Erlang is two fold:
>
> 1. I want to have fun making the project.
> 2. The site has to be extremely scalable. Although I might not move it
> out to large scale work I want it to be able to handle it.
>
> Other requirements I have are as follows:
>
> 1. Will be hosted on a linux machine. Does anyone have any notes on
> performance of Erlang and also Yaws on different 'brands' of linux.

At most a linux that has epoll support in the kernel and of course an
erlang VM compiled for using epoll. Though I never bothered caring
about this myself so I dont know if this is enabled by default in
current releases?

I doubt you will have to care about this as early in the project, it
is something you can "correct" later without any change to your code.

> 2. I will want to have use a distributed filesystem instead of
> database to access data. Any good recommendations in this context.
> Database will still probably be used to hold pointers to specific data
> but data will be in filesystem.

This seems rather non-trivial to get right. Is it a way to work around
blob-inefficiency in mnesia that has been made infamous the last
weeks?

First I dont know if the blob inefficy is unfair criticism or founded
in actual bottlenecks everyone storing more than atomic types in their
tables end up with. I'm certainly not afraid of using mnesia until it
breaks rather than avoiding it altogheter in fear that i might have to
work around it later. Of course, my needs are limited to a few hundred
ten(s)-of-kilobyte documents.

Anyway, if your files are large and you want them replicated over a
fraction of the nodes in a cluster you might want to look at

http://en.wikipedia.org/wiki/Distributed_hash_table

Joe was in #erlang arising my interest for distributed hashes a week
or two ago. I wonder if he finished any thoughts on it.  There are
some very interesting research in peer to peer filesharing, i suggest
you look around for what scientists have published.