Selective Network Traffic Throttling in Virtual Data Centers
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In virtual data center environments, the deployment and management of network clusters require significant manual configuration and result in heavy network traffic, with WAN bandwidth congestion being a costly issue due to the need for manual negotiation of protocol parameters and the pausing of all initiators during congestion, even when only a few applications or compute nodes are the primary cause.
Innovation Solution
A management server selects and programs network nodes to eliminate manual provisioning steps by determining logical paths and protocol parameters, and selectively throttles traffic from identified contributors to congestion using out-of-band connections and flow table programming in switches.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If manual configuration and negotiation of protocol parameters is performed for each host and storage, then deployment flexibility and customization are improved, but deployment time and network traffic increase significantly
Solution Approach 1:
The system enables self-service deployment by allowing hosts and storage devices to automatically negotiate and configure protocol parameters without administrator intervention. The fabric login and transport protocol negotiation occur automatically when devices connect to the network, eliminating manual configuration steps while maintaining deployment flexibility.
Solution Approach 2:
The system performs preliminary configuration actions by pre-establishing fabric login credentials and transport protocol parameters during device initialization. This preliminary setup enables automatic negotiation and configuration when devices first connect to the network, significantly reducing deployment time while maintaining adaptability.
2Reliability
If pause frames are transmitted to all initiators during network congestion, then network congestion is controlled, but all data traffic is paused including non-contributing applications
Solution Approach 1:
The system applies local quality by implementing selective congestion control that targets specific initiators or applications causing congestion rather than applying uniform pause frames to all initiators. The congestion control mechanism identifies and throttles only the contributing sources while allowing non-contributing applications to maintain their data traffic flow.
Solution Approach 2:
The system segments the congestion control approach by dividing initiators into contributing and non-contributing groups based on their traffic patterns and congestion impact. This segmentation enables differential treatment where only contributing initiators receive pause frames or throttling, while non-contributing initiators continue uninterrupted data transfer.
3Reliability
If all initiators have their data traffic paused during congestion, then network stability is maintained, but bandwidth is wasted and application performance deteriorates
Solution Approach 1:
The system implements local quality by applying congestion control measures selectively to only those initiators contributing to network congestion. Non-contributing initiators continue their data transfers without interruption, thereby maintaining network stability while avoiding bandwidth wastage and preserving application performance.
Solution Approach 2:
The system converts the harmful effect of congestion into a beneficial selective throttling mechanism. By identifying and targeting only the initiators causing congestion, the system transforms the potential harm of widespread traffic pausing into a benefit where network stability is maintained through precise control, and bandwidth is preserved for non-contributing applications.
Data Source
AI summary
Exemplary methods, apparatuses, and systems a management server receiving, from a switch, an indication of network congestion at the switch. The indication includes one or more contributors to the network congestion. In response to the indication of network congestion, the management server identifies one or more hosts running the identified contributor(s) and transmits an instruction to the host(s) to restrict network traffic originated by the contributor(s). For example, the restriction may include a reduction in the contributor's I/O queue depth. If the reduction in traffic does not remedy the congestion, the management server may receive a further indication of congestion and further instruct the host(s) to restrict identified contributor network traffic. For example, the contributors may have their I/O queue depth reduced exponentially until congestion is relieved. Upon receiving an indication that the congestion is relieved, the management server instructs the host(s) to remove the restriction on network traffic.