Services Oriented Architecture with PHP and MySQL

Services Oriented Architecture with PHP

and MySQLJoe Stump, Lead Architect, Digg

Introductions

I’m Joe Stump, Lead Architect for Digg.

Reduce HTTP Requests

Bundle JavaScript and CSS Use sprites for images Reduce images / outside objects

This talk doesn’t cover MySQL details. The point of this discussion is how to minimize MySQL’s impact on your company’s data layer.

Old habits die hard

Data requests are sequential Data requests are blocking Tightly coupled Scaling is not abstracted

Services Oriented Architecture (SOA)

Data is requested from a service Data requests are ran in parallel Data requests are asynchronous Data layer is loosely coupled Scalability is abstracted

What are my options?

Running requests over HTTP New York Times’ DBSlayer Danga’s Gearman DIY

HTTP w/ PHP

1. Group requests for data at the top2. Open a socket for each request

1. Sockets must be non-blocking2. Make sure to TCP_NODELAY

3. Use __get() to block for results4. See Services_Digg_Request

Grouping Data Requests

Opening the Sockets

Opening the Sockets

Returning Data

HTTP Conclusions

HTTP is widely supported in all languages Easy to get up and running Lots of options for servers / tuning Lots of overhead in protocol Apache has a lot of overhead

DBSlayer

Small HTTP daemon written in C Uses JSON for communications Connection pooling Load balancing and failover Tightly coupled to MySQL (for now) Tightly coupled to SQL No intelligence

Gearman

Highly scalable queuing system Simple/Efficient binary protocol Jobs can return results (e.g. data) Sets of jobs are ran in parallel Queue can scale linearly PHP, Perl, Python, Ruby, C clients Poorly documented Not very “robust”

Do It Yourself

Highly customized solutions (Flickr) Extremely efficient for custom cases Customize your protocols Requires more resources

What goes in the Service layer?

Smart caching strategies Data mapping and distribution Intelligent grouping of data results Partitioning logic

DO WANT!

Intelligently group data into endpoints Version your endpoints Bundle and group requests

EPIC FAIL!

Don’t just send SQL over for translation Hundreds of teeny tiny endpoints Not running SOA requests in parallel

ResourcesServices_Digghttp://pear.php.net/package/Services_Digg

Danga’s Gearmanhttp://www.danga.com/gearman/

Net_Gearmanhttp://code.google.com/p/netgearman/

New York Timesʼ DBSlayerhttp://code.nytimes.com/projects/dbslayer

Questions?!

Contact/Flame Me

Joe [email protected]://joestump.net