Partitioned serialized caching and delivery of large files
Some embodiments provide partitioned serialized caching and delivery of large sized content and files. Some embodiments partition requests for large sized content into segment requests with each segment request identifying a different byte range of the requested content. Each segment request is hashed to identify a particular server from a set of servers tasked with caching and delivering a different segment of the requested content. In this manner, no single server caches or delivers the entirety of large sized content. The segment requests are distributed serially across the set of servers so that the segments are passed in order, wherein the serial distribution involves handing-off the requesting user's connection serially to each server of the set of server in the order with which the set of servers deliver the content segments.
1. A method comprising:
receiving, from a client device at a request management server, a client request directed to a file, wherein the client request does not specify a byte range;
generating, from the client request at the request management server, a first subrequest directed to a first byte range of the file, and a second subrequest directed to a second byte range of the file that does not overlap with the first byte range, wherein the first subrequest and the second subrequest comprise at least one of a different header or Uniform Resource Locator (URL) than the client request received from the client device;
distributing (i) the first subrequest, instead of the client request, from the request management server to a first server of a set of servers in a common site with the request management server, and the second subrequest, instead of the client request, from the request management server to a second server of the set of servers;
serving the first byte range of the file to the client device from the first server in response to said distributing of the first subrequest; and
serving the second byte range of the file to the client device from the second server in response to said distributing of the second subrequest.
2. The method of claim 1 further comprising determining that a size of the file exceeds a size threshold.
3. The method of claim 2 further comprising caching the first byte range of the file at the first server, and the second byte range of the file at the second server in response to the size of the file exceeding the size threshold, wherein the first server and the second server preserve storage by caching mutually exclusive and different parts of the file.
4. The method of claim 2 further comprising selecting the first server to serve the first byte range based on the size of the file exceeding the size threshold and a result of hashing an identifier identifying the file and the first byte range, and selecting the second server to serve the second byte range based on the size of the file exceeding the size threshold and a result of hashing the identifier identifying the file and the second byte range.
5. The method of claim 1 , wherein the first subrequest specifies the first byte range in a range request HyperText Transfer Protocol (HTTP) header field of a first request message, and the second subrequest specifies the second byte range in the range request HTTP header field of a second request message.
6. The method of claim 1 further comprising routing the client request from the client device to the request management server based on a hash of an identifier identifying the file, routing the first subrequest from the request management server to the first server based on a hash of the identifier and the first byte range, and routing the second subrequest from the request management server to the second server based on a hash of the identifier and the second byte range.
7. The method of claim 1 further comprising obtaining a size of the file in response to receiving the client request from the client device.
8. The method of claim 7 , wherein obtaining the size comprises issuing a head request from the request management to an origin server that stores said file, and receiving the size without contents of the file from the origin server in response to the head request.
9. The method of claim 7 , wherein obtaining the size comprises performing a lookup of the size based on an identifier identifying the file in the client request, and identifying the size based on the lookup.
10. The method of claim 7 further comprising partitioning the client request into at least the first and second subrequests based on the size of the file exceeding a size threshold.
11. The method of claim 10 , wherein said partitioning comprises setting the first byte range to span a first number of bytes equal to a segment stripe size, and setting the second byte range to span a second number of bytes that immediately follow the first number of bytes and that equal the segment stripe size.
12. The method of claim 1 , wherein said generating comprises appending the first byte range to a URL of the first subrequest, and appending the second byte range to a URL of the second subrequest.
13. A method comprising:
receiving, from a client device, a client request directed to a first byte range of a file;
generating, from the client request, a first subrequest with a second byte range starting at a first byte of the first byte range and ending at a last byte that is before a last byte of the first byte range, wherein the first subrequest comprises at least one of a different header or Uniform Resource Locator (URL) than the client request received from the client device;
generating, from the client request, a second subrequest with a third byte range starting at a byte after the last byte of the second byte range and ending at the last byte of the first byte range, wherein the second subrequest comprises at least one of a different header or URL than the client request and the first subrequest;
distributing (i) the first subrequest, instead of the client request, to a first server of a set of servers, and (ii) the second subrequest, instead of the client request, to a second server of the set of servers;
serving the second byte range of the file to the client device directly from the first server in response to said distributing of the first subrequest; and
serving the third byte range of the file to the client device directly from the second server in response to said distributing of the second subrequest, wherein the second server is different than the first server.
14. The method of claim 13 further comprising determining that the first byte range is larger than a segment stripe size for cached segments of the particular file, and partitioning the request into at least the first subrequest and the second subrequest based on said determining.
15. The method of claim 13 further comprising routing the first subrequest to the first server based on hashing an identifier and the second byte range of the first subrequest, and routing the second subrequest to the second server based on hashing the identifier and the third byte range of the second subrequest.
16. The method of claim 13 , wherein said generating the first subrequest comprises setting a first byte of the second byte range to coincide with the first byte of the first byte range, and setting a last byte of the second byte range to span a segment stripe size from the first byte of the second byte range.
17. The method of claim 13 further comprising caching the second byte range of the file at the first server, and caching the third byte range of the file at the second server.
18. The method of claim 13 , wherein the first subrequest specifies the second byte range in a range request HyperText Transfer Protocol (HTTP) header field, and the second subrequest specifies the third byte range in the range request HTTP header field.
19. The method of claim 13 , wherein the first subrequest specifies the second byte range as parameters of a URL directed to said file.
20. A system comprising:
a first server comprising storage caching a first byte range of a file;
a second server comprising storage caching a second non-overlapping byte range of the file;
a third server comprising:
one or more processors configured to:
receive a client request that is directed to the file from a client device;
generate, from the client request, a first subrequest directed to the first byte range of the file, and a second subrequest directed to the second byte range of the file that does not overlap with the first byte range, wherein the first subrequest and the second subrequest comprise at least one of a different header or Uniform Resource Locator (URL) than the client request received from the client device;
route, to the first server, the first subrequest directed to the file and the first byte range of the file instead of the client request, wherein the first server serves the first byte range of the file directly to the client device in response to receiving the first subrequest;
route, to the second server, the second subrequest directed to the file and the second byte range of the file instead of the client request, and wherein the second server directly serves the second byte range of the file to the client device in response to receiving the second subrequest.