To minimize the time it takes a user to download a webpage, web content is often...

Transcribed Image Text:

To minimize the time it takes a user to download a webpage, web content is often repli- cated at multiple locations. This problem explores two basic approaches for replicating web content: Web Caching and CDNs. CDN-based approaches rely on a content distri- bution network to serve content from multiple locations; there are (at least) three ways to implement the CDN distribution model. Thus, we have four different options at our disposal: a) Web Caching: A local network routes all HTTP requests through a Web Cache. The cache first checks if the requested page is already cached, in which case the proxy re- turns the cached page to the user. If the page is not cached, the proxy requests the page from the site, caches the result, and returns it to the user. b) CDN using DNS: When the authoritative server for the site receives a DNS request, it returns one of many IP addresses, each of which corresponds to a CDN server hosting the web content. c) CDN using IP Anycast: DNS requests for the site always return the same, single IP address. Multiple CDN servers (at various different locations) that host the web content advertise this same IP address. d) CDN based on rewriting media URLs: All requests for the base page of a site are served from a single server hosted by that site. The site's server re-writes the URLs for media content (e.g., images) to be URLs corresponding to CDN servers, so that media content is served from the CDN. Haobo Yu's new web startup has gone viral, and he wants to purchase services from a CDN in order to minimize the load on your servers. List one of the options, or 'None' if appropriate. 1. Which approach allows the finest-grained control over which users are directed to which servers? In particular, which approach allows a site to dictate exactly how many users are sent to each server, and to instantaneously re-balance load between servers? To minimize the time it takes a user to download a webpage, web content is often repli- cated at multiple locations. This problem explores two basic approaches for replicating web content: Web Caching and CDNs. CDN-based approaches rely on a content distri- bution network to serve content from multiple locations; there are (at least) three ways to implement the CDN distribution model. Thus, we have four different options at our disposal: a) Web Caching: A local network routes all HTTP requests through a Web Cache. The cache first checks if the requested page is already cached, in which case the proxy re- turns the cached page to the user. If the page is not cached, the proxy requests the page from the site, caches the result, and returns it to the user. b) CDN using DNS: When the authoritative server for the site receives a DNS request, it returns one of many IP addresses, each of which corresponds to a CDN server hosting the web content. c) CDN using IP Anycast: DNS requests for the site always return the same, single IP address. Multiple CDN servers (at various different locations) that host the web content advertise this same IP address. d) CDN based on rewriting media URLs: All requests for the base page of a site are served from a single server hosted by that site. The site's server re-writes the URLs for media content (e.g., images) to be URLs corresponding to CDN servers, so that media content is served from the CDN. Haobo Yu's new web startup has gone viral, and he wants to purchase services from a CDN in order to minimize the load on your servers. List one of the options, or 'None' if appropriate. 1. Which approach allows the finest-grained control over which users are directed to which servers? In particular, which approach allows a site to dictate exactly how many users are sent to each server, and to instantaneously re-balance load between servers?