Dynamic queue management of network traffic
AssigneeCisco Technology, Inc.
InventorsJerome Henry, Robert Edgar Barton
Life of this patent
Estimating the term…
Abstract
Techniques and systems described herein relate to network system queue management and dynamic real-time re-allocation of resources to prevent oversubscription and packet loss due to oversubscription. The techniques and systems enable monitoring of traffic and initial identification of queues at risk for oversubscription based on a rate of change of traffic load on the queue in advance of oversubscription occurring. After identifying a queue at risk for oversubscription, an Extended Berkeley Packet Filter or other similar component performs a likelihood determination using predictive algorithm techniques to identify a likelihood of oversubscription in the near future and re-allocates to parallel queues for efficient and loss-free use of the queues.
Claims20
A method for dynamic packet reallocation, the method comprising:
- monitoring traffic through a network system using an application hosted in a user space of a network device;
- determining, in response to monitoring the traffic, a load of a first queue of the network system is approaching full capacity;
- calculating, using a packet filter of a virtual machine of the network system, a likelihood score indicative of a probability that the load will exceed a capacity limit within a predetermined time period;
- and
- re-balancing the first queue with at least one second queue of the network system in response to the likelihood score exceeding a predetermined threshold.
A system comprising:
- one or more processors;
- and
- one or more non-transitory computer-readable media storing computer-executable instructions that, when executed by the one or more processors, cause the one or more processors to perform operations comprising:
- monitor traffic through a network system using an application hosted in a user space of a network device;
- determine, in response to monitoring the traffic, a load of a first queue of the network system is approaching full capacity;
- calculate, using a packet filter of a virtual machine of the network system, a likelihood score indicative of a probability that the load will exceed a capacity limit within a predetermined time period;
- and
- re-balance the first queue with at least one second queue of the network system in response to the likelihood score exceeding a predetermined threshold.
A controller device comprising:
- one or more processors;
- and
- one or more non-transitory computer-readable media storing computer-readable instructions that, when executed by one or more processors, cause the one or more processors to:
- monitor traffic through a network system using an application hosted in a user space of a network device;
- determine, in response to monitoring the traffic, a load of a first queue of the network system is approaching full capacity;
- calculate, using a packet filter of a virtual machine of the network system, a likelihood score indicative of a probability that the load will exceed a capacity limit within a predetermined time period;
- and
- re-balance the first queue with at least one second queue of the network system in response to the likelihood score exceeding a predetermined threshold.
Dependencies are inferred from the claim text.