Load sharing among loosely coupled or independent video servers
Methods, systems, and apparatuses can provide a pooling architecture for video servers and network resources. The pooling architecture can be based upon allocating content object copies to two or more nodes, intelligently selecting a primary streaming resource and a failover streaming resource, and identifying and compensating for exception asymmetries in system resource demand.
1. A computer-implemented method for allocating resources, comprising:
providing a plurality of network video servers operating as nodes within a network and are operable to serve streaming video content to playback devices;
allocating a content object copy to two or more network video servers from the plurality of network video servers, the two or more network video servers comprising an object group associated with the content object copy, wherein a content object copy is allocated to and stored at each of the two or more network video servers before identifying an exceptional asymmetry with regard to a demand for the content object copy;
receiving a request for the content object copy;
responsive to the request for the content object copy selecting a primary node to source the stream and a failover node to provisionally reserve failover capacity for redundancy, the primary and failover nodes being selected from the object group; and
identifying an exceptional asymmetry in capacity, wherein the exceptional asymmetry is identified with regard to the demand for the content object copy by determining that one or more increases in the demand for the content object copy have consumed at least a threshold share of total streaming capacity;
allocating one or more additional content object copies to compensate for the exceptional asymmetry; and
allocating a plurality of content objects to the plurality of network video servers so as to intersect a number of different object groups, wherein the intersecting groups can be uniquely associated with a network video server and one of the other network video servers, thereby providing that a set of object groups associated with any network video server in the plurality of network video servers overlap exactly one group belonging to another network video server, while each set of object groups associated with a network video server are unique.
2. The computer-implemented method of claim 1 , further comprising:
allocating the content object copy to two or more network video servers based on a best-fit content allocation model.
3. The computer-implemented method of claim 1 , further comprising:
allocating the content object copy to two or more network video servers based on a random content allocation model.
4. The computer-implemented method of claim 1 , further comprising:
allocating the content object copy to two or more network video servers based on hybrid content allocation model using a best-fit content allocation component and a random content allocation component.
5. The computer-implemented method of claim 1 , wherein a collection of object copies of one network video server are allocated so as to be different than collections of object copies associated with any other network video server.
6. The computer-implemented method of claim 1 , further comprising selecting a primary network video server to source a new stream is based on maintaining a balanced maximum streaming load across the plurality of network video servers.
7. The computer-implemented method of claim 6 , wherein a worst case scenario for a load associated with a network video server is matched to a relative streaming capacity associated with the network video server.
8. The computer-implemented method of claim 7 , wherein the worst case scenario for the load is identified based on a current streaming load and a maximum failover load associated with the network video server.
9. A resource pooling system, comprising:
a plurality of network video servers operating as nodes within a network, the plurality of network video servers being operable to serve video content to playback devices;
at least one processor, communicatively coupled to the network, configured to provide a content manager operable to allocate a content object copy to two or more network video servers from the plurality of network video servers, the two or more network video servers comprising an object group associated with the content object copy, wherein a content object copy is allocated to each of the two or more network video servers before identifying an exceptional asymmetry with regard to a demand for the content object copy; and
wherein, responsive to a request for the content object copy, the content manager is further operable to select a primary node to source the stream and a failover node to provisionally reserve failover capacity for the stream, the primary and failover nodes being selected from the object group;
wherein the content manager is further operable to identify an exceptional asymmetry in capacity and to allocate one or more additional content object copies to compensate for the exceptional asymmetry, wherein the exceptional asymmetry is identified with regard to the demand for the content object copy by determining that one or more increases in the demand for the content object copy have consumed at least a threshold share of total streaming capacity; and
wherein the content manager is operable to allocate a plurality of content objects to the network video servers so as to intersect each content object to a number of different object groups, and further wherein the intersecting groups can be uniquely associated with a network video server and one of the other network video servers, thereby providing that a set of object groups associated with any network video server in the plurality overlap exactly one group belonging to another node, while each set of object groups associated with a network video server are unique.
10. The system of claim 9 , wherein the content manager is operable to allocate the content object copy to the two or more network video servers based on a best-fit content allocation model.
11. The system of claim 9 , wherein the content manager is operable to allocate the content object copy to two or more network video servers based on a random content allocation model.
12. The system of claim 9 , wherein the content manager is operable to allocate the content object copy to the two or more network video servers based on hybrid content allocation model using a best-fit content allocation component and a random content allocation component.
13. The system of claim 9 , wherein a collection of object copies of one network video server from the plurality of network video servers are allocated so as to be different than collections of object copies associated with any other network video server.
14. The system of claim 9 , wherein the content manager is operable to select a primary network video server to source a new stream is based on maintaining a balanced maximum streaming load across the plurality of network video servers.
15. The system of claim 9 , wherein a worst case scenario for a load associated with a plurality of video servers is matched to a relative streaming capacity associated with the network video server.
16. The system of claim 9 , wherein the worst case scenario for the load is identified based on a current streaming load and a maximum failover load associated with the network video server.