Method and apparatus for low latency data center network

Patent metadata

Field	Value
Publication number	US-10873529-B2
Application number	US-201715829319-A
Country	US
Kind code	B2
Filing date	Dec 1, 2017
Priority date	Dec 1, 2017
Publication date	Dec 22, 2020
Grant date	Dec 22, 2020

Abstract

Methods and apparatus for determining and adjusting traffic rate limits to reduce queuing delays and latency in a data center network. In various embodiments, a network controller receives network topology information and traffic data relating to flows of the network. The traffic data includes traffic matrix information, flow path information, and pause frame distribution information. Based on this information, the network controller identifies server network node interfaces that are the source of a pause frame. Flows associated with such network node interfaces are also identified and ranked based on traffic volume. For ranked flows meeting predetermined criteria, reduced traffic rate limits are calculated and sent to associated server network node interfaces. In some embodiments, the reduced traffic rate limits are incrementally recovered when normal traffic resumes on a paused link. In further embodiments, traffic rate limits are also adjusted for flows of highly utilized and underutilized network node interfaces.

First claim

What is claimed is:

1. A computer-implemented method for reducing latency in a network having a plurality of servers, the method comprising:
receiving traffic data relating to flows of the network, the traffic data including traffic matrix information, flow path information, and pause frame distribution information;
identifying, based on network topology information and the pause frame distribution information, one or more network node interfaces of the plurality of servers that are a source of a pause frame;
identifying flows in the traffic matrix information which pass through the one or more network node interfaces;
determining, based on the traffic data and predetermined criteria, a reduced traffic rate limit for each of a plurality of the identified flows, the determining further comprising
ranking the identified flows based on respective flow traffic volume, and
decreasing a current traffic rate limit for each of a ranked flows in which the flow traffic volume exceeds the predetermined criteria;

identifying, based on the network topology information and the traffic data, one or more additional network node interfaces having a utilization rate greater than a predetermined threshold;
reducing an associated traffic rate limit for flows passing through the one or more additional network node interfaces with the utilization rate greater than the predetermined threshold;
determining adjusted traffic rate limits by incrementally increasing the reduced traffic rate limit for flows of a network node interface, of the one or more network node interfaces, identified as no longer a source of pause frame based on the reduced traffic rate limit;
determining further adjusted traffic rate limits by incrementally decreasing the reduced traffic rate limit for the identified flows of the network node interface when the network node interface is the source of a pause frame based on the adjusted traffic rate limits; and
determining further adjusted traffic rate limits by incrementally increasing the reduced traffic rate limit for the identified flows of the network node interface when the network node interface is not the source of a pause frame based on the adjusted traffic rate limits.

2. The computer-implemented method of claim 1 , wherein the predetermined criteria comprises:
ranking the flows of an identified network node interface based on respective flow traffic volume values of the flows; and
identifying one or more of the ranked flows having a cumulative traffic volume that constitutes a predetermined portion of a total flow traffic volume of the identified network node interface.

3. The computer-implemented method of claim 2 , wherein decreasing the current traffic rate limit by a percentage basis.

4. The computer-implemented method of claim 1 further comprising:
receiving updated pause frame distribution information;
identifying, based on the network topology information and the updated pause frame distribution information, the network node interface, of the one or more network node interfaces, that is no longer the source of the pause frame based on the reduced traffic rate limit; and
sending the adjusted traffic rate limits for application to the network node interface.

5. The computer-implemented method of claim 4 further comprising:
receiving further updated pause frame distribution information;
determining whether the further updated pause frame distribution information indicates that the network node interface is the source of the pause frame based on the adjusted traffic rate limits; and
sending the further adjusted traffic rate limits for application to the network node interface.

6. The computer-implemented method of claim 1 further comprising:
identifying, based on the traffic data, additional flows of the additional network node interfaces;
determining additional reduced traffic rate limit for one or more of the additional flows of the additional node interfaces; and
sending the additional reduced traffic rate limit for application to the additional network node interfaces.

7. The computer-implemented method of claim 1 further comprising:
identifying, based on the network topology information and the traffic data, the additional network node interfaces of the plurality of servers having the utilisation rate less than the predetermined threshold;
identifying, based on the traffic data, additional flows of the additional network node interfaces, wherein each network node interface of an additional flow is an additional network node interface;
determining increased traffic rate limits for the additional flows of the additional network node interfaces; and
sending the increased traffic rate limits for application to the additional network node interfaces.

8. The computer-implemented method of claim 1 , further comprising:
identifying, based on the traffic data, initial flows of the plurality of servers;
determining, based on the traffic data, an initial traffic rate limit for each of the initial flows; and
sending the initial rate limits for application to the plurality of servers.

9. The computer-implemented method of claim 1 , wherein the network is a data center network.

10. The computer-implemented method of claim 1 , wherein the method is repeated on a periodic basis.

11. A device comprising:
a memory storage comprising instructions;
at least one network interface; and
one or more processors in communication with the memory storage, wherein the one or more processors execute the instructions to:
receive, via the at least one network interface, traffic data relating to flows of the network, the traffic data including traffic matrix information, flow path information, and pause frame distribution information;
identify, based on network topology information and the pause frame distribution information, one or more network node interfaces of a plurality of servers that are a source of a pause frame;
identify flows in the traffic matrix information which pass through the one or more network node interfaces;
determine, based on the traffic data and predetermined criteria, a reduced traffic rate limit for each of a plurality of the identified flows, and further comprising
ranking the identified flows based on respective flow traffic volume, and
decreasing a current traffic rate limit for each of the ranked flows in which the flow traffic volume exceeds the predetermined criteria;

identify, based on the network topology information and the traffic data, one or more additional network node interfaces having a utilization rate greater than a predetermined threshold;
reduce an associated traffic rate limit for flows passing through the one or more additional network node interfaces with the utilization rate greater than the predetermined threshold;
determine adjusted traffic rate limits by incrementally increasing the reduced traffic rate limit for flows of a network node interface, of the one or more network node interfaces, identified as no longer a source of pause frame based on the reduced traffic rate limit;
determine further adjusted traffic rate limits by incrementally decreasing the reduced traffic rate limit for the identified flows of the network node interface when the network node interface is a source of the pause frame based on the adjusted traffic rate limits; and
determine further adjusted traffic rate limits by incrementally increasing the reduced traffic rate limit for the identified flows of the network node interface when the network node interface is not a source of the pause frame based on the adjusted traffic rate limits.

12. The device of claim 11 , wherein the predetermined criteria compr

Assignees

• Futurewei Technologies Inc

Inventors

• Liu Fangping
• Li Zhenjiang
• Avci Serhat Nazim

Classifications

• H04L47/12 Primary

  Avoiding congestion; Recovering from congestion · CPC title

• H04L47/25 Primary

  with rate being modified by the source upon detecting a change of network conditions · CPC title

• H04L45/02

  Topology update or discovery · CPC title