Gray's blog

Well this blog isn’t anyways.

I will be posting more / deleting this post when I have something useful to say. Right now most projects are under wraps, so it’s hard to say a lot.

As for Axcid 2.0 updates, it has been decided that the entire backend portion of the site (the API, the search tools, the database, the index, the relational DB) will all be powered by a Java servlet backend.

This means data processing will be faster / more extensible & clean. The API will include an insert API in 2.0 for private use (Internal crawling). More types of data will be indexed (bitrate, id3 tags, file size).

The server situation will be changing, built on 2 dedicated servers, one will be a workhorse, and one will be a datastore.

The workhorse will be the server used for streaming content, running large indexing operations & holding large amounts of data.

The indexing/logic machine will be used for the larger, more complex data manipulation systems. It will also manage searching/api responses/site control.

Song checking & large operations that require multiple machines will be operated under clusters, using a popular Java distributed operations API.

This is just some of what is planned / in development for Axcid 2.0. The wait will be well worth it, taking a music database to a proper, music search engine.

After a while of running this music DB, and noticing lag spikes and slowdowns due to poor, old coding, I made a decision. It’s rewrite time.

Now I know I’ve been saying that forever, but this time Axcid changes itself from a database to a search engine. I’ve engineered a new storage format / searchable database that can store music data, literally 50 times as fast as before. Queries that once took 50 seconds now take 20 milliseconds.

The new system will be much more scalable. Work starting with multiple servers, splitting the ‘powerhouse’ load (streaming, content checking, ‘heavy lifting’), from the database load (searching, indexing, database handling).

The XML api’s will also be re engineered. They will use a better / more standard format.

There will also be many new features, including album indexing, better art retrieval, users / sessions & playlists. The software version will be rebuilt to a more solid, complete application.

We’re moving out of the prototype days, and into the real software days. The beginning was fun, but it’s time to buckle down and launch a real music service.