« Do you know of some... | Main | The Father's Day... »
20080613 Friday June 13, 2008

LinkedIn's Engineering Blog LinkedIn Blog Have you been curious about LinkedIn's architecture or how they're using Grails and Rails? If so, you might be interested in LinkedIn's Engineering Blog. Over the past couple of weeks, a few Engineers have starting writing about our architecture, OpenSocial, RailsConf, YUI, Grails and OSGi. Below is a complete listing of Engineering posts.

If there are topics you'd like to see us blog about, please let me know. I've somehow landed in the role of Editor for the Engineering Blog, so I should be able to hook you up if I can find an engineer to blog about what you're interested in.

On a related note, Rob Getzschman's entry LinkedIn discovers the truth about Cannes is quite entertaining. Highly recommended. Posted in Java at Jun 13 2008, 08:30:19 AM MDT 9 Comments

Comments:

I've looped through the linkedIn stuff and I've one question. According the architecure prensentation they use Tomcat and Jetty. Those are the webcontainer parts in J2EE. Why aren't they deployed on top of some application server? (Did I mis a part!) What's the reason for only using a webcontainer in large scale projects?

Posted by koenvda on June 17, 2008 at 06:03 AM MDT #

Backend services are deployed in multiple instance of Jetty. The frontend is deployed in multiple versions of Tomcat. Most super high traffic sites like LinkedIn, Amazon and eBay use a stateless architecture to handle the load. Most application servers (and Java EE) are inclined to lean towards stateful and clustered servers. Ours is much more disconnected and componentized - where more servers can be added to serve more instances of a particular service. Of course, most web applications won't need to scale like LinkedIn does, so traditional Java EE will likely work fine for them.

Posted by Matt Raible on June 17, 2008 at 03:47 PM MDT #

Hi Matt, I have another question. Did Hibernate's exclusion from the technology stack have anything to do with its inability to scale? Would love to hear your thoughts on it.

Posted by Allan Ang on June 17, 2008 at 10:25 PM MDT #

> Did Hibernate's exclusion from the technology stack have anything to do with its inability to scale?

No, the technology stack was simply created before Hibernate was that popular. We're using Hibernate with our Grails applications. It's possible we'll use it in the future. In the meantime, straight JDBC works pretty well for us. Why fix something that isn't broken?

Posted by Matt Raible on June 18, 2008 at 10:41 AM MDT #

Hi Matt, I have another question based on the post by koenvda. How do you guys in linkedIn manage state?

Posted by Allan Ang on July 08, 2008 at 11:34 AM MDT #

I was wondering, how does caching work with pure jdbc. In hibernate we cache objects, when you use pure jdbc what do you cache ?

Posted by sid on December 16, 2009 at 08:30 PM MST #

I was wondering, how does caching work with pure jdbc. In hibernate we cache objects, when you use pure jdbc what do you cache ?

Posted by sid on December 16, 2009 at 08:30 PM MST #

I was wondering, how does caching work with pure jdbc. In hibernate we cache objects, when you use pure jdbc what do you cache ?

Posted by sid on December 16, 2009 at 08:30 PM MST #

I was wondering, how does caching work with pure jdbc. In hibernate we cache objects, when you use pure jdbc what do you cache ?

Posted by sid on December 16, 2009 at 08:31 PM MST #

Post a Comment:
  • HTML Syntax: Allowed