Systems and methods for configuring networks
The disclosed computer-implemented method may include (i) generating a data center constraint model by placing a constraint on a total amount of ingress or egress traffic a service expects from each respective data center of multiple data centers, (ii) filtering a set of traffic matrices that indicate points in the data center constraint model by comparing the set of traffic matrices against cut sets of a network topology that indicate network failures to create a tractable set of dominating traffic matrices, (iii) obtaining physical network resources to implement a cross-layer network upgrade architecture that satisfies the tractable set of dominating traffic matrices, and (iv) allocating the physical network resources across the multiple data centers according to the cross-layer network upgrade architecture such that a capacity level of the multiple data centers is increased while satisfying the data center constraint model. Various other methods, systems, and computer-readable media are also disclosed.
1. A computer-implemented method comprising:
generating a data center constraint model by placing a constraint on a total amount of ingress or egress traffic a service expects from each respective data center of multiple data centers;
filtering a set of traffic matrices that indicate points in the data center constraint model by comparing the set of traffic matrices against cut sets of a network topology that indicate network failures to create a tractable set of dominating traffic matrices;
obtaining physical network resources to implement a cross-layer network upgrade architecture that satisfies the tractable set of dominating traffic matrices; and
allocating the physical network resources across the multiple data centers according to the cross-layer network upgrade architecture such that a capacity level of the multiple data centers is increased while satisfying the data center constraint model.
2. The computer-implemented method of claim 1 , wherein the traffic matrices indicate a volume of traffic that is forecast between each pair of data centers in the multiple data centers.
3. The computer-implemented method of claim 1 , wherein the network failures comprise at least two of singular fiber cuts, dual submarine link failures, and repeated failures.
4. The computer-implemented method of claim 1 , wherein filtering the set of traffic matrices is performed through integer linear programming.
5. The computer-implemented method of claim 1 , wherein the data center constraint model indicates a convex polytope.
6. The computer-implemented method of claim 1 , wherein the physical network resources comprise fiber and equipment.
7. The computer-implemented method of claim 1 , wherein filtering the set of traffic matrices selects traffic matrices that are predicted to drive requirements for the physical network resources.
8. The computer-implemented method of claim 1 , wherein ensuring that the cross-layer network upgrade architecture satisfies the tractable set of dominating traffic matrices guarantees, according to a predefined probability threshold, that the cross-layer network upgrade architecture will also satisfy a remainder of the set of traffic matrices.
9. The computer-implemented method of claim 1 , wherein a number of traffic matrices in the tractable set of dominating traffic matrices is minimized.
10. The computer-implemented method of claim 1 , wherein comparing the set of traffic matrices against the cut sets of the network topology that indicate network failures comprises selecting a set of traffic matrices that generates a maximum level of traffic over each graph cut of the cut sets.
11. The computer-implemented method of claim 1 , wherein generating the cross-layer network upgrade architecture comprises sizing a link across each graph cut of the cut sets.
12. The computer-implemented method of claim 11 , wherein each graph cut of the cut sets forms a substantially straight line.
13. The computer-implemented method of claim 12 , wherein each graph cut of the cut sets forms a straight line.
14. The computer-implemented method of claim 1 , wherein generating the cross-layer network upgrade architecture is performed according to a cost optimization model.
15. The computer-implemented method of claim 14 , wherein the cost optimization model generates the cross-layer network upgrade architecture through integer linear programming.
16. The computer-implemented method of claim 14 , wherein the cost optimization model accounts for physical constraints in terms of at least three of:
fiber count;
space and power;
spectrum consumption on fiber; or
maximum capacity constraints on leased waves.
17. The computer-implemented method of claim 14 , wherein the cost optimization model accounts for cost of at least three of:
fiber procurement;
optical or Internet Protocol hardware;
operational cost;
space and power; or
provisioning cost.
18. The computer implemented method of claim 1 , wherein the cross-layer network upgrade architecture is generated according to a hose-based computation rather than a pipe-based computation.
19. A system comprising:
a processor; and
a memory comprising instructions that, when executed, cause the processor to:
generate a data center constraint model by placing a constraint on a total amount of ingress or egress traffic a service expects from each respective data center of multiple data centers;
filter a set of traffic matrices that indicate points in the data center constraint model by comparing the set of traffic matrices against cut sets of a network topology that indicate network failures to create a tractable set of dominating traffic matrices;
facilitate the obtaining of physical network resources to implement a cross-layer network upgrade architecture that satisfies the tractable set of dominating traffic matrices; and
allocate the physical network resources across the multiple data centers according to the cross-layer network upgrade architecture such that a capacity level of the multiple data centers is increased while satisfying the data center constraint model.
20. A non-transitory computer-readable medium comprising instructions that, when executed, cause a computing device to:
generate a data center constraint model by placing a constraint on a total amount of ingress or egress traffic a service expects from each respective data center of multiple data centers;
filter a set of traffic matrices that indicate points in the data center constraint model by comparing the set of traffic matrices against cut sets of a network topology that indicate network failures to create a tractable set of dominating traffic matrices;
facilitate the obtaining of physical network resources to implement a cross-layer network upgrade architecture that satisfies the tractable set of dominating traffic matrices; and
allocate the physical network resources across the multiple data centers according to the cross-layer network upgrade architecture such that a capacity level of the multiple data centers is increased while satisfying the data center constraint model.