CN116455822A - Self-adaptive load balancing method based on heterogeneous flow in data center network - Google Patents

Self-adaptive load balancing method based on heterogeneous flow in data center network Download PDF

Info

Publication number
CN116455822A
CN116455822A CN202310239698.1A CN202310239698A CN116455822A CN 116455822 A CN116455822 A CN 116455822A CN 202310239698 A CN202310239698 A CN 202310239698A CN 116455822 A CN116455822 A CN 116455822A
Authority
CN
China
Prior art keywords
flow
stream
data
short
long
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202310239698.1A
Other languages
Chinese (zh)
Inventor
王进
黄莉莎
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Changsha University of Science and Technology
Original Assignee
Changsha University of Science and Technology
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Changsha University of Science and Technology filed Critical Changsha University of Science and Technology
Priority to CN202310239698.1A priority Critical patent/CN116455822A/en
Publication of CN116455822A publication Critical patent/CN116455822A/en
Pending legal-status Critical Current

Links

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L47/00—Traffic control in data switching networks
    • H04L47/10—Flow control; Congestion control
    • H04L47/12—Avoiding congestion; Recovering from congestion
    • H04L47/125—Avoiding congestion; Recovering from congestion by balancing the load, e.g. traffic engineering
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L47/00—Traffic control in data switching networks
    • H04L47/10—Flow control; Congestion control
    • H04L47/19—Flow control; Congestion control at layers above the network layer
    • H04L47/193—Flow control; Congestion control at layers above the network layer at the transport layer, e.g. TCP related
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L47/00—Traffic control in data switching networks
    • H04L47/10—Flow control; Congestion control
    • H04L47/24—Traffic characterised by specific attributes, e.g. priority or QoS
    • H04L47/2441—Traffic characterised by specific attributes, e.g. priority or QoS relying on flow classification, e.g. using integrated services [IntServ]
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L47/00—Traffic control in data switching networks
    • H04L47/10—Flow control; Congestion control
    • H04L47/24—Traffic characterised by specific attributes, e.g. priority or QoS
    • H04L47/2483—Traffic characterised by specific attributes, e.g. priority or QoS involving identification of individual flows
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D30/00—Reducing energy consumption in communication networks
    • Y02D30/50—Reducing energy consumption in communication networks in wire-line communication networks, e.g. low power modes or reduced link rate

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Data Exchanges In Wide-Area Networks (AREA)

Abstract

The invention provides a self-adaptive load balancing method based on heterogeneous traffic, which aims at the field of data center network load balancing and comprises the following steps: s1, an exchanger receives a data stream sent by a transmitter in a data center network; s2, the switch judges whether the data stream is a delay sensitive stream (short stream) or a throughput sensitive stream (long stream) or a stream delivered by the best effort, and a first judging result is obtained; s3, the switch determines a routing path corresponding to the data flow according to the first judging result; s4, the switch forwards the data stream to a receiver in a data center network through the routing path. According to the technical scheme, the problem that in the prior art, the flow completion time of the short flow and the throughput rate of the long flow cannot be balanced due to resource competition between the short flow and the long flow can be solved, the link load condition in a network is perceived in real time, and the balance of network flow is realized rapidly.

Description

Self-adaptive load balancing method based on heterogeneous flow in data center network
Technical Field
The application relates to the technical field of communication, in particular to a self-adaptive load balancing method based on heterogeneous traffic in a data center network.
Background
Modern data centers typically host a variety of applications, including mainly delay-sensitive applications (e.g., search, rpc, and gaming) and throughput-sensitive applications (e.g., video, big data analysis, and VM migration). Delay-sensitive applications typically generate a large number of short streams, each of which transmits a small amount of data (less than 100 kb), requiring a Data Center Network (DCN) to provide a low queuing delay. Throughput sensitive applications may generate a small number of long streams. Each long stream has a large amount of data (over 100 KB) and requires a long-lasting throughput during data transmission. These heterogeneous streams are mixed in DCN. In order to provide better quality of service and user experience, DCNs need to meet their performance requirements at the same time. However, this challenge is still elusive.
Fortunately, in recent progress, there is a very promising approach to address the challenges described above. The topology of a DCN (e.g., leaf-ridge topology) typically has multiple parallel end-to-end paths. An efficient load balancing mechanism is deployed on the DCN, and all the parallel paths can be fully utilized, so that a higher network halving bandwidth is provided, and the network transmission performance is improved. However, the proliferation of end-to-end bandwidth is not expected to bring significant performance improvement due to the inefficiency of the path selection strategy of existing data center load balancing schemes.
Most of the existing load balancing schemes cannot avoid resource contention between short and long flows. The mixed stream collides with other streams on the same path, affecting the final result due to their opposing demands. Long-stream packets consume a large amount of switching buffer space to achieve higher utilization, while short-stream packets must experience head-of-line congestion and significant queuing delays. The large queuing delay greatly increases the completion time of short flows that are prone to miss deadlines defined by the Service Level Agreement (SLA). Reducing the occupancy of the switch buffer, while improving the performance of the short stream, cannot guarantee that the link bandwidth is fully utilized, resulting in a loss of throughput for the long stream.
In recent years, some schemes are used for routing heterogeneous traffic in different ways, so that granularity and routing paths of long flows are effectively limited according to distribution of short flows, and the long flows and the short flows are isolated, but the rerouting mode causes that the proportion of the paths occupied by the short flows is too large, thereby influencing routing results of the long flows and further reducing throughput rate of the long flows.
Disclosure of Invention
In order to solve the technical problems, the application provides a heterogeneous flow-based self-adaptive load balancing method in a data center network, which can solve the problem that in the prior art, resource competition between short flows and long flows causes unbalanced flow completion time of the short flows and throughput rate of the long flows, sense link load conditions in the network in real time, and rapidly realize balance of network flows.
The technical scheme provided by the application is as follows:
the application provides a heterogeneous flow-based self-adaptive load balancing method in a data center network, which is applied to a switch in the data center network and comprises the following steps of: s1, an exchanger receives a data stream sent by a transmitter in a data center network; s2, the switch judges whether the data stream is a delay sensitive stream (short stream) or a throughput sensitive stream (long stream) or a stream delivered by the best effort, and a first judging result is obtained; s3, the switch determines a routing path corresponding to the data flow according to the first judging result; s4, the switch forwards the data stream to a receiver in a data center network through the routing path.
Further, in a preferred mode of the present application, the S3 includes:
if the first judging result is that the data flow is a short flow, determining a path with the shortest queue length in all paths at present as the routing path corresponding to the short flow, and routing to the path with flow granularity to avoid the disorder problem;
if the first judging result is that the data flow is the flow delivered in the best effort, determining a path with the longest queue length in all paths at present as the routing path corresponding to the flow delivered in the best effort, and routing to the path with packet granularity to ensure the throughput of long and short flows;
if the first judging result is that the data flow is long flow, randomly routing the data flow to an available path, judging whether a short flow exists in a current path queue, obtaining a second judging result, and determining the switching granularity of the long flow according to the second judging result;
further, in a preferred mode of the present application, determining the rerouting granularity corresponding to the long flow according to the second determination result includes:
and when the second judging result is that the current path does not contain the short flow, the long flow is selected to be rerouted with the flow granularity.
According to the traffic load balancing method, traffic is divided into three categories at the switch, and corresponding routing paths are selected according to traffic classification, so that the problem of resource competition between short flows and long flows is solved; the short stream is routed with stream granularity, so that the short stream is not disordered, the throughput rate of the flow can be guaranteed by the self-adaptive granularity of the long stream, and meanwhile, the throughput rates of the long stream and the short stream can be guaranteed simultaneously by selecting the path with highest delay by the packet granularity of the stream delivered in the best effort; because the flow characteristics in the data center network are that about 80% of the flow is provided by about 20% of the throughput-sensitive long flows, about 80% of the delay-sensitive short flows are provided by about 20% of the flow, the method and the device adapt to the flow characteristics by adopting the modes of short flow granularity and long flow self-adaptive switching granularity, so as to achieve the purposes of solving heterogeneous flow resource competition, short flow low-time delay and long flow high throughput, solve the problem that the performance requirements of long and short flows cannot be met at the same time in the prior art, sense the link load condition in the network in real time and realize the balance of network flow rapidly.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, it being obvious that the drawings in the following description are only some embodiments of the invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
Fig. 1 is a timing chart of an adaptive load balancing method based on heterogeneous traffic according to an embodiment of the present application;
fig. 2 is a flowchart of an adaptive load balancing method based on heterogeneous traffic according to an embodiment of the present application.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments of the present invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Specifically stated in connection with the embodiments, fig. 1 is a timing chart of an adaptive load balancing method based on heterogeneous traffic provided in the embodiments of the present application, as shown in fig. 1, and compared with the prior art, the adaptive load balancing method based on heterogeneous traffic provided in the present application includes the following steps:
s1, a transmitter transmits a data stream to a switch;
s2, the exchanger receives the data stream;
s3, the exchanger judges that the data stream is a long stream, a short stream or a stream exchanged by the best effort;
s41, if the switch judges that the data flow is a short flow, selecting a shortest route with the length of the queue as a route path of the short flow;
s42, if the switch judges that the data flow is the flow delivered in the best effort, selecting the longest route of the queue length as the route path of the flow delivered in the best effort;
s43, if the exchanger judges that the data stream is a long stream, judging whether a short stream exists in a queue of the current path;
s431, if there is short flow in the queue of the present route, the long flow uses the data packet as granularity as the rerouting decision;
s432, if no short flow exists in the queue of the current path, taking a flowlet as granularity by a long flow as a rerouting decision;
s5, the exchanger forwards the data stream to the receiver;
s6, the receiver receives the data stream;
s7, the receiver submits the data stream to a TCP layer;
based on the above method flow, the following describes in detail the main steps:
in S3, the long stream is a data stream with a data size greater than or equal to a preset value, the short stream is a data stream with a data size less than the preset value, and the stream delivered in the best effort is a background stream, which can be judged by the switch;
in S41, the short stream is routed to the path with the shortest queue length according to the stream granularity, so that the short stream is ensured not to be disordered, and the stream completion time is improved;
in S42, the best effort delivered flow adopts packet granularity routing to the path with the longest queue length, and since this class of flow has no performance requirement, only needs to be successfully delivered to the receiver, the routing scheme can protect the transmission efficiency of short flows and long flows;
in S43 we maintain a short flow identification table at each path of the switch, when a short flow is newly identified to be present on that path, the count bit is incremented by 1 and the valid bit is changed to 1; when a short stream is identified to be sent, the counting bit is decremented by 1; only when the count bit in the table becomes 0 will the valid bit be updated to 0. The long stream is rerouted with packet granularity in the path with the valid bit of 1, and is rerouted with flowlet granularity in the path with the valid bit of 0;
in S431, if there is a short flow in the queue of the current path, then the long flow reroutes with packet granularity, giving the path where the short flow is located more space, ensuring low delay of the short flow, meanwhile, since most of the time period of the short flow is distributed on most paths, such routing decision also ensures that the long flow has routing, and ensures throughput rate of the long flow;
if there is no short flow in the queue of the current path, long flows are routed with the flowlet granularity, and the transmission path is selected again for the traffic only when the flowlet time-out exceeds 500 μs, i.e. the path is switched only when the time interval between the data packet and the last data packet exceeds 500 μs (the flowlet time-out supports the maximum traffic rate of about 500 μs. We use this value in the design) to ensure that there is no congestion in the path.
In view of the foregoing, in the data center network according to the embodiment of the present application, the adaptive load balancing method based on heterogeneous traffic is based on load balancing of a switch, and on one hand, a rerouting decision is determined for a long flow by determining whether there is a short flow in a queue of the switch; the short flow is routed with the flow granularity, so that the short flow is not disordered, the throughput rate of the flow can be guaranteed by the self-adaptive granularity of the long flow, and the throughput rates of the long flow and the short flow are guaranteed simultaneously by the routing with the packet granularity of the flow delivered in the best effort; the method adapts to the flow characteristics in a mode of short flow granularity and long flow self-adaptive switching granularity so as to achieve the purposes of short flow low time delay and long flow high throughput; on the other hand, the method can be independent of a network stack of a host, can immediately serve all flows once deployed, can sense the link load condition in a network in real time, and can quickly realize the balance of network flows on a path connected with an outlet port of a switch; and the method can be implemented on the basis of a commercial switch without customizing the switch, and can solve the problem that the flow completion time of the short flow and the throughput rate of the long flow cannot be balanced due to resource competition between the short flow and the long flow in the prior art, sense the link load condition in the network in real time and quickly realize the balance of network flow.
In this specification, each embodiment is described in a related manner, and identical and similar parts of each embodiment are all referred to each other, and each embodiment mainly describes differences from other embodiments. In particular, for system embodiments, since they are substantially similar to method embodiments, the description is relatively simple, as relevant to see a section of the description of method embodiments.
The foregoing description is only of the preferred embodiments of the present invention and is not intended to limit the scope of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention are included in the protection scope of the present invention.

Claims (4)

1. An adaptive load balancing method based on heterogeneous flow in a data center network is applied to a switch in the data center network, and comprises the following steps:
s1, receiving a data stream sent by a transmitter in a data center network;
s2, judging that the data stream is a delay-sensitive short stream, a throughput-sensitive long stream or a stream delivered by the best effort, and obtaining a first judgment result, wherein the long stream is a data stream with the data size being larger than or equal to a preset value, the short stream is a data stream with the data size being smaller than the preset value, the stream delivered by the best effort is a background stream, and the switch can judge by itself;
s3, determining a routing path corresponding to the data flow according to the first judging result;
s4, forwarding the data stream to a receiver in a data center network through the routing path.
2. The method according to claim 1, wherein S3 comprises:
if the first judging result is that the data flow is a short flow, determining a path with the shortest queue length in all paths at present as the routing path corresponding to the short flow, and routing to the path with flow granularity to avoid the disorder problem;
if the first judging result is that the data flow is the flow delivered in the best effort, determining a path with the longest queue length in all paths at present as the routing path corresponding to the flow delivered in the best effort, and routing to the path with packet granularity to ensure the throughput of long and short flows;
and if the first judging result is that the data flow is long flow, randomly routing the data flow to an available path, judging whether a short flow exists in a current path queue, obtaining a second judging result, and determining the rerouting switching granularity of the long flow according to the second judging result.
3. The method according to claim 2, wherein determining the reroute switching granularity corresponding to the long flow according to the second determination result includes:
and when the second judging result is that the current path does not contain the short flow, the long flow is selected to be rerouted with the flow granularity.
4. A traffic load balancing system in a data center network, comprising:
a transmitter in the data center network for transmitting the data stream to a switch in the data center network;
the switch is configured to receive a data stream sent by the sender in a data center network, determine that the data stream is a long stream, a short stream or a stream delivered by best effort, and obtain a first determination result, where the long stream is a data stream with a data size greater than or equal to a preset value, the short stream is a data stream with a data size less than the preset value, and the stream delivered by best effort is a background stream, where the switch can determine by itself; if the first judging result is that the data flow is long, randomly routing the data flow to an available path, judging whether a short flow exists in a current path queue, and obtaining a second judging result; determining a routing path and a rerouting switching granularity of the data flow according to the first judging result and the second judging result, and forwarding the data flow to a receiver in a data center network through the routing path;
a receiver in the data center network for receiving the data stream forwarded by the switch in the data center network.
CN202310239698.1A 2023-03-10 2023-03-10 Self-adaptive load balancing method based on heterogeneous flow in data center network Pending CN116455822A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310239698.1A CN116455822A (en) 2023-03-10 2023-03-10 Self-adaptive load balancing method based on heterogeneous flow in data center network

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310239698.1A CN116455822A (en) 2023-03-10 2023-03-10 Self-adaptive load balancing method based on heterogeneous flow in data center network

Publications (1)

Publication Number Publication Date
CN116455822A true CN116455822A (en) 2023-07-18

Family

ID=87132822

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310239698.1A Pending CN116455822A (en) 2023-03-10 2023-03-10 Self-adaptive load balancing method based on heterogeneous flow in data center network

Country Status (1)

Country Link
CN (1) CN116455822A (en)

Similar Documents

Publication Publication Date Title
US10243865B2 (en) Combined hardware/software forwarding mechanism and method
KR100977651B1 (en) Method and apparatus for network congestion control
US8553543B2 (en) Traffic shaping method and device
Prabhavat et al. On load distribution over multipath networks
US8467295B2 (en) System and methods for distributed quality of service enforcement
EP1810466B1 (en) Directional and priority based flow control between nodes
KR100656509B1 (en) Congestion avoidance method for video service bandwidth
US7830889B1 (en) Systems for scheduling the transmission of data in a network device
CN110351187B (en) Load balancing method for path switching granularity self-adaption in data center network
US9197570B2 (en) Congestion control in packet switches
US20030169690A1 (en) System and method for separating communication traffic
EP1476994B1 (en) Multiplexing of managed and unmanaged traffic flows over a multi-star network
JP2005236669A (en) Method and device for controlling communication quality
CN116455822A (en) Self-adaptive load balancing method based on heterogeneous flow in data center network
Bonald et al. Scheduling network traffic
Alanazi et al. Ml-ecn: Multi-level ecn marking for fair datacenter traffic forwarding
CN115396357B (en) Traffic load balancing method and system in data center network
CN116708302A (en) Load balancing system based on self-adaptive switching granularity in data center network
Alanazi et al. AQ-CAR: adaptive-queueing congestion-aware routing for datacenter traffic forwarding
CN117176658A (en) Flowlet load balancing method for data center network receiving end driving transmission protocol
Li et al. An Application-Driven-based Network Resource Control Method
JP2003333088A (en) Packet transfer system
CN118233381A (en) Protocol agnostic cognitive congestion control
Prabhavat A study on traffic distribution models over multipath networks
KR20130022781A (en) Rio and rio-dc method for qos guarantee of the assured service in differentiated services through network within a vessel

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination