Photo Caching on Social Sites
6:34 am in infrastructure management, miscellaneous, monitoring by Matt Jenkins
I read the bbc news story today – which reports that a group at Cambridge University found that the direct URL to uploaded photos on various sites continued to exist 30 days after they deleted them.
Is this news? These sites have several hosting challenges, one key area is caching.
Facebook (for example) stores millions of photos, and needs to make all of them fast for customers in different countries and cheap to serve. Using a CDN (Content Delivery Network) like Akamai, Limelight, or EdgeCast is the only way to achieve this.
A CDN has servers in many different countries, and many different networks. These servers are mainly large caching boxes, a request comes in to them, they either serve the request, or go back to the main site and then cache the result for future queries. The cache time can be set to any length of time, from 24 hours to forever (or the most useful “until my alloted memory runs out”). If this is how a site has their CDN setup, then deleting every instance of a cached image would be a hugely inefficient task.
The social network sites get a lot of complaints about privacy, and they have a responsibility to the users to keep them secure, but the primary responsibility for privacy on the net is with the individual.