| 摘要 |
In an exemplary embodiment, a computer-implemented method includes providing a mapping between a set of two or more IP addresses and a set of two more servers. A weight recommendation for the servers is received, based at least in part on server data related to current states of the servers, where the weight recommendation represents a goal for distributing a plurality of transactions among the servers. The mapping is modified based at least in part on the weight recommendation. A transaction is received, directed toward a first IP address of the two or more IP addresses. The transaction is routed to a first server, of the two or more servers, associated with the first IP address according to the modified mapping. |
| 主权项 |
1. A computer-implemented method comprising:
providing a first mapping from two or more IP addresses to two more servers, the first mapping being embodied in a virtual network layer modifiable by a workload advisor based on a weight recommendation; receiving from a load balancer a first plurality of transactions directed to the two or more IP addresses; receiving the weight recommendation for the two or more servers based at least in part on server data related to current states of the two or more servers, wherein the weight recommendation defines a goal for weighting transactions among the two or more servers; determining dynamically, by a computer processor of the workload advisor, a second mapping from the two or more IP addresses to the two or more servers, the second mapping being based on the weight recommendation, absent communication between the load balancer and the workload advisor; transitioning dynamically from the first mapping to the second mapping, responsive to receiving the weight recommendation; wherein the transitioning comprises modifying how many of the two or more IP addresses are mapped to each of the two or more servers, wherein after the transitioning, how many of the two or more IP addresses that are mapped to each of the two or more servers corresponds to the weight recommendation, and wherein at least two IP addresses map to a first server according to the second mapping; performing a traffic reduction by turning off traffic to a first IP address of the two or more IP addresses, wherein the traffic reduction is performed through an application programming interface of the load balancer, and wherein the traffic reduction is performed based on the weight recommendation; receiving a second plurality of transactions directed to the two or more IP addresses; and routing the second plurality of transactions to the two or more servers through the virtual network layer according to the second mapping and the traffic reduction, wherein the routing comprises:
assigning the second plurality of transactions to the two or more IP addresses using a round-robin distribution, independent of the weight recommendation;directing a first transaction, of the second plurality of transactions, to a first IP address to which the first transaction is assigned in the round-robin distribution; andidentifying a destination server, of the two or more servers, that is associated with the first IP address in the second mapping, wherein the destination server for the first transaction is determined based on the round robin distribution and based on the second mapping. |
