Here is a speculation:
Like youtube and youku that are customized CDNs (traditional direct purchase as's, blue sin, ali CDN this provide a standard service pattern, do not seem to provide distributed source station [like distributed storage, multiple nodes collection content stored in a common but not simple node completely copy], customized edge computing services such as [is to put some don't need to fully centralized service migration to] where closer to the user, personal observations is that you can provide one or several domain name to CDNs providers, they put these domain names as the source station,), user uploaded content may be directly stored in a layer of CDN node, and the nodes could form a source of distributed storage, and other post-processing class computing tasks may be distributed in a layer of CDNs to edge of computing nodes, otherwise why so fast,
CDNs's main aim is to unblock the most original network, reduce the backbone transmission network traffic, is optimizing the route cache + + protocol optimization, compression, but seemed to assume that more content from basic content providers (youtube) itself, rather than from millions of users, that is to say, now not only to solve the problem of content providers to distribute, to solve its content quickly collect and even some of the user needs to consume a certain amount of calculation of custom processing problem, in fact, youtube is a representative of this kind of application, all similar "large-scale users upload - processing - the process of distribution" of the application should can apply this sample, therefore, would like to ask for the application of this current CDNs whether can solve the problem, whether can pass the CDN + cloud computing to solve the problem? thank you
CodePudding user response:
This problem is very strange, clearly his thought of method, try this method can determine not line not line?