TWI641268B - System of optical switch used in data center network - Google Patents

System of optical switch used in data center network Download PDF

Info

Publication number
TWI641268B
TWI641268B TW106135329A TW106135329A TWI641268B TW I641268 B TWI641268 B TW I641268B TW 106135329 A TW106135329 A TW 106135329A TW 106135329 A TW106135329 A TW 106135329A TW I641268 B TWI641268 B TW I641268B
Authority
TW
Taiwan
Prior art keywords
traffic
optical
buffer device
intelligent scheduling
end buffer
Prior art date
Application number
TW106135329A
Other languages
Chinese (zh)
Other versions
TW201918078A (en
Inventor
鄭玉鉅
蕭毅
陳伯濤
Original Assignee
中華電信股份有限公司
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 中華電信股份有限公司 filed Critical 中華電信股份有限公司
Priority to TW106135329A priority Critical patent/TWI641268B/en
Application granted granted Critical
Publication of TWI641268B publication Critical patent/TWI641268B/en
Publication of TW201918078A publication Critical patent/TW201918078A/en

Links

Landscapes

  • Data Exchanges In Wide-Area Networks (AREA)
  • Optical Communication System (AREA)

Abstract

本發明揭露一種運用於資料中心網路之光交換系統,其包含前端緩衝器設備、中央智能排程控制器以及光流量交換機,該中央智能排程控制器連接該前端緩衝器設備與該光流量交換機,其中,前端緩衝器設備接收來自一伺服器群的流量訊務,該中央智能排程控制器依據該前端緩衝器設備所接收之流量訊務,判斷出該流量訊務具衝突情況時,啟動智能排程演算法以管控該前端緩衝器設備之運作,並建立虛擬光路徑,而該光流量交換機依據該中央智能排程控制器之管控,利用該中央智能排程控制器建立之該虛擬光路徑執行虛擬光電路交換,俾達到該流量訊務之快速交換。 The invention discloses an optical switching system applied to a data center network, comprising a front end buffer device, a central intelligent scheduling controller and an optical flow switch, wherein the central intelligent scheduling controller connects the front end buffer device and the optical flow a switch, wherein the front-end buffer device receives traffic traffic from a server group, and the central intelligent scheduling controller determines, according to the traffic information received by the front-end buffer device, that the traffic signal has a conflict condition. A smart scheduling algorithm is started to control the operation of the front-end buffer device, and a virtual optical path is established, and the optical traffic switch establishes the virtual state by using the central intelligent scheduling controller according to the control of the central intelligent scheduling controller. The optical path performs virtual optical circuit switching to achieve fast switching of the traffic.

Description

運用於資料中心網路之光交換系統 Optical switching system for data center networks

本發明係關於一種網路系統,特別的是,係一種運用於資料中心網路上之光交換系統。 The present invention relates to a network system, and more particularly to an optical switching system for use in a data center network.

中華民國專利號I552536揭露一種光資料中心網路系統以及光交換器,其所提的光交換器為商用化的光選擇交換器(Wavelength Selective Switch,WSS)所構成,將光資料中心的網路系統分三層架構來實現,包括多個第一層光交換器、多個第二層光交換器以及多個第三層光交換器,其中,多個第一層光交換器透過帶狀光纖(ribbon fiber)互相連接形成一群組(pod),而多個第二層光交換器透過帶狀光纖互相連接形成一巨群組(macro pod),且每一個第二層光交換器並與一個群組中的所有第一層光交換器連接,最後,多個第三層光交換器也是透過帶狀光纖互相連接,且每一個第三層光交換器並與一個巨群組中的所有第二層光交換器連接。此專利主要是針對光資料中心網路系統,利用三層金字塔架構來實現。 The Republic of China Patent No. I552536 discloses an optical data center network system and an optical switch, and the proposed optical switch is composed of a commercially available Wavelength Selective Switch (WSS), which is a network of optical data centers. The system is implemented in a three-layer architecture, including multiple first layer optical switches, multiple second layer optical switches, and multiple third layer optical switches, wherein multiple first layer optical switches pass through the ribbon optical fibers. The ribbon fibers are interconnected to form a pod, and the plurality of second layer optical switches are interconnected by the ribbon fibers to form a macro pod, and each of the second layer optical switches is coupled to All the first layer optical switches in a group are connected. Finally, multiple third layer optical switches are also connected to each other through the ribbon optical fibers, and each of the third layer optical switches is associated with all of the macro groups. The second layer of optical switch is connected. This patent is mainly for the optical data center network system, which is realized by the three-layer pyramid structure.

另外,美國專利公開號為US 20140205292,其發明名 稱為OPTICAL PACKET SWITCHING SYSTEM,為一種光封包交換機的設計,其所提的光緩衝器是一般的Fiber Delay Line的設計,僅做長時間的光延遲設計,並無訊務flow智慧型排程的光緩衝設計,且也未針對光流量交換機上的光緩衝器需求而設計。 In addition, U.S. Patent Publication No. US 20140205292, the name of the invention Known as OPTICAL PACKET SWITCHING SYSTEM, it is a design of optical packet switch. The optical buffer is a general Fiber Delay Line design. It only uses long-term optical delay design, and there is no traffic flow intelligent scheduling. The optical buffer design is also not designed for optical buffer requirements on optical flow switches.

由上可知,現有習用方式無法有效提升光交換機的交換總量(Throughput),因而如何找出一種網路系統,特別是,可運用於資料中心網路上,並且具設備簡化優點且可提升交換總量之光交換系統,此將成為本技術領域人員努力追求之目標 It can be seen from the above that the existing usage method cannot effectively improve the total amount of switching of the optical switch, so how to find a network system, in particular, can be applied to the data center network, and has the advantages of simplifying the device and improving the total exchange. Optical switching system, which will become the goal pursued by those skilled in the art

本發明提出一種光交換系統,可運用於資料中心網路上,主要是利用智慧型排程機制技術加上可程式化的光流量交換機的系統設計,使原來在資料中心的光流量交換機之交換埠減少,但配合智慧型排程機制下,卻使整體交換總量(Throughput)較原光交換機進一步增加,期盼達到交換埠數量減少,但整體系統交換總量增加之目的。 The invention provides an optical switching system, which can be applied to a data center network, mainly by using a smart scheduling mechanism technology and a system design of a programmable optical flow switch, so that the optical traffic switch originally in the data center is exchanged. Reduced, but with the intelligent scheduling mechanism, but the total exchange total (Throughput) is further increased than the original optical switch, looking forward to the reduction of the number of exchanges, but the overall system exchange total increase.

本發明係提出一種運用於資料中心網路之光交換系統,其包含:前端緩衝器設備,係接收來自一伺服器群的流量訊務;中央智能排程控制器,係連接該前端緩衝器設備,用以依據該前端緩衝器設備所接收之流量訊務,判斷該流量訊務是否具衝突情況,以於判斷出該流量訊務具衝突情況時,啟動智能排程演算法以管控該前端緩衝器設備之運作,並建立虛擬光路徑;以及光流量交換機,係連接 該中央智能排程控制器與該前端緩衝器設備,用以依據該中央智能排程控制器之管控,利用該中央智能排程控制器建立之該虛擬光路徑執行虛擬光電路交換,俾達到該流量訊務之快速交換。 The present invention provides an optical switching system for a data center network, comprising: a front-end buffer device for receiving traffic traffic from a server group; and a central intelligent scheduling controller for connecting the front-end buffer device And determining, according to the traffic information received by the front-end buffer device, whether the traffic signal has a conflict condition, so as to determine that the traffic signal has a conflict condition, start an intelligent scheduling algorithm to control the front-end buffer The operation of the device and establish a virtual light path; and the optical flow switch, the connection The central intelligent scheduling controller and the front-end buffer device are configured to perform virtual optical circuit switching by using the virtual optical path established by the central intelligent scheduling controller according to the control of the central intelligent scheduling controller. Fast exchange of traffic messages.

於一實施例中,該前端緩衝器設備為2n*n之可程式化的L2網路交換機。另外,對應該前端緩衝器設備之設計,該光流量交換機為n*n之核心光流量交換機。 In one embodiment, the front end buffer device is a 2n*n programmable L2 network switch. In addition, in response to the design of the front-end buffer device, the optical traffic switch is a n*n core optical traffic switch.

於另一實施例中,該中央智能排程控制器判斷該流量訊務未有衝突情況時,傳送直通指令至該前端緩衝器設備,以令該前端緩衝器設備將該流量訊務直接傳遞至該光流量交換機。 In another embodiment, the central intelligent scheduling controller determines that the traffic signal does not have a conflict condition, and sends a through command to the front end buffer device, so that the front end buffer device directly transmits the traffic message to the The optical flow switch.

於又一實施例中,該智能排程演算法包括檢查產生衝突之流量訊務的優先權,以令高優先權者直接進入該光流量交換機,以及令低優先權者進入該前端緩衝器設備以進行緩衝處理。 In yet another embodiment, the intelligent scheduling algorithm includes checking the priority of the conflicting traffic traffic to enable the high priority person to directly enter the optical traffic switch and to enable the low priority person to enter the front end buffer device. For buffering.

該智能排程演算法更包括於該流量訊務之優先權相同時,檢查該流量訊務之級別長度,以令級別大之流量訊務具有較高的優先權,俾優先從該前端緩衝器設備進入該光流量交換機,而令級別小之流量訊務繼續進入該前端緩衝器設備以進行緩衝處理。 The intelligent scheduling algorithm further includes checking the level length of the traffic signal when the priority of the traffic message is the same, so that the traffic with a higher level has a higher priority, and the priority buffer is preferentially obtained from the front buffer. The device enters the optical traffic switch, and the traffic with a small level continues to enter the front-end buffer device for buffer processing.

該智能排程演算法更包括於該流量訊務之級別長度相同時,檢查時間排程,以依據時間軸的FIFO,將先進入FIFO排程之流量訊務優先處理。 The intelligent scheduling algorithm further includes checking the time schedule when the level of the traffic signal is the same, so that the traffic FIFO that first enters the FIFO schedule is preferentially processed according to the FIFO of the time axis.

該智能排程演算法更包括於該流量訊務進入該FIFO 排程時,每隔一段時間將該流量訊務在該FIFO排程的優先權增加。 The intelligent scheduling algorithm is further included in the traffic message entering the FIFO. During scheduling, the priority of the traffic message in the FIFO schedule is increased at regular intervals.

於再一實施例中,該中央智能排程控制器係預先建立該虛擬光路徑,以透過時間交換機制操控該光流量交換機進行該虛擬光電路交換。 In still another embodiment, the central intelligent scheduling controller pre-establishes the virtual optical path to control the optical optical switch to perform the virtual optical circuit switching through a time switch mechanism.

於又另一實施例中,該光流量交換機接收該前端緩衝器設備所傳送之流量訊務,俾依據該中央智能排程控制器預先設定的光交換路徑,進行該流量訊務的交換與傳送。 In still another embodiment, the optical traffic switch receives the traffic information transmitted by the front-end buffer device, and performs the exchange and transmission of the traffic information according to the optical switching path preset by the central intelligent scheduling controller. .

本發明提出一個包括可程式化光流量交換機的系統設計,配合智慧型排程機制技術,在前端緩衝器設備上進行中央智能排程機制的管控,將最適當的訊務flow經過智慧型排程機制下,送往光流量交換機進行光交換傳送,因此,在此一系統設計下,配合智慧型排程機制,可程式化的光流量交換機之交換埠數量可減少,但整體交換總量較原先的光交換機卻進一步增加。綜上,在光流量交換機研究上,未有人提出可程式化的光流量交換機之設計,且透過光流量交換機並配合智能排程機制的管控,無論在交換埠數量減少數量上,或是整體系統交換總量增加上,皆有顯著的進步,可進一步達成綠能減碳的終極目的。 The invention provides a system design including a programmable optical flow switch, and cooperates with a smart scheduling mechanism technology to perform central intelligent scheduling control on the front-end buffer device, and the most appropriate traffic flow is intelligently scheduled. Under the mechanism, it is sent to the optical traffic switch for optical switching transmission. Therefore, under the design of this system, with the intelligent scheduling mechanism, the number of exchangeable switches of the programmable optical traffic switch can be reduced, but the total exchange amount is larger than the original one. The optical switch has been further increased. In summary, in the research of optical flow switches, no one has proposed the design of a programmable optical flow switch, and through the optical flow switch and the management of the intelligent scheduling mechanism, whether in the number of exchanges, or the overall system Significant progress has been made in the increase in the total amount of exchange, which can further achieve the ultimate goal of green energy reduction.

1‧‧‧運用於資料中心網路之光交換系統 1‧‧‧ Optical switching system for data center networks

10‧‧‧前端緩衝器設備 10‧‧‧ front-end buffer device

11‧‧‧中央智能排程控制器 11‧‧‧Central Intelligent Scheduling Controller

12‧‧‧光流量交換機 12‧‧‧ Optical Flow Switch

20‧‧‧2n*2n網路交換機 20‧‧‧n*2n network switch

22‧‧‧2n*2n光流量交換機 22‧‧‧2n*2n optical flow switch

100‧‧‧伺服器群 100‧‧‧Server group

S41~S45‧‧‧流程 S41~S45‧‧‧ Process

第1圖為本發明之運用於資料中心網路之光交換系統的示意系統架構;第2A和2B圖為本發明使用128*128的核心光流量交換機以及傳統256*256的核心光流量交換機的架構示意 圖;第3圖為本發明之光交換系統在不同路徑數下整體光流量交換機之交換總量(Throughput)的比較圖;以及第4圖為本發明之運用於資料中心網路之光交換系統中智能排程演算法的檢測流程。 1 is a schematic system architecture of an optical switching system applied to a data center network according to the present invention; FIGS. 2A and 2B are diagrams showing a core optical traffic switch using 128*128 and a core optical traffic switch of a conventional 256*256 according to the present invention; Schematic diagram Figure 3 is a comparison diagram of the total amount of exchanges of the overall optical flow switches of the optical switching system of the present invention under different path numbers; and Figure 4 is an optical switching system for the data center network of the present invention. The detection process of the intelligent scheduling algorithm.

以下藉由特定的具體實施形態說明本發明之技術內容,熟悉此技藝之人士可由本說明書所揭示之內容輕易地瞭解本發明之優點與功效。然本發明亦可藉由其他不同的具體實施形態加以施行或應用。 The technical contents of the present invention are described below by way of specific embodiments, and those skilled in the art can easily understand the advantages and effects of the present invention from the contents disclosed in the present specification. The invention may be embodied or applied by other different embodiments.

本發明為運用於資料中心網路上包括可程式化光流量交換機的系統設計,主要是利用可程式化光流量交換機應用在資料中心上,利用智慧型排程機制技術加上可程式化光流量交換機的系統設計,使得前端緩衝器設備、光流量交換機、中央智能排程控制器構成一光交換系統。 The present invention is designed for use in a data center network including a programmable optical flow switch, mainly using a programmable optical flow switch applied to a data center, utilizing intelligent scheduling mechanism technology and a programmable optical flow switch. The system design makes the front-end buffer device, the optical flow switch, and the central intelligent scheduling controller form an optical switching system.

請參閱第1圖,其所示為本發明之運用於資料中心網路之光交換系統的示意系統架構。如圖所示,本發明之一種運用於資料中心網路之光交換系統1包括:前端緩衝器設備10、中央智能排程控制器11以及光流量交換機12。 Please refer to FIG. 1, which shows a schematic system architecture of an optical switching system for a data center network of the present invention. As shown, an optical switching system 1 for a data center network of the present invention includes a front end buffer device 10, a central intelligent scheduling controller 11, and an optical flow switch 12.

前端緩衝器設備10係接收來自一伺服器群100的流量訊務,由於本案是應用於光交換系統,故流量訊務即指光流量訊務,但並不以此為限。 The front-end buffer device 10 receives the traffic information from a server group 100. Since the present invention is applied to the optical switching system, the traffic information refers to the optical traffic signal, but is not limited thereto.

中央智能排程控制器11係連接該前端緩衝器設備10,該中央智能排程控制器11依據該前端緩衝器設備10 所接收之流量訊務,判斷出該流量訊務具衝突情況時,啟動智能排程演算法以管控該前端緩衝器設備10之運作,並建立虛擬光路徑。 The central intelligent scheduling controller 11 is connected to the front end buffer device 10, and the central intelligent scheduling controller 11 is based on the front end buffer device 10 When the received traffic message determines that the traffic message has a conflict condition, the intelligent scheduling algorithm is started to control the operation of the front-end buffer device 10 and establish a virtual light path.

光流量交換機12係連接該中央智能排程控制器11與該前端緩衝器設備10,該光流量交換機12依據該中央智能排程控制器11之管控,基於該虛擬光路徑執行虛擬光電路交換,以達到該流量訊務之快速交換。 The optical flow switch 12 is connected to the central intelligent scheduling controller 11 and the front end buffer device 10, and the optical flow switch 12 performs virtual optical circuit switching based on the virtual optical path according to the control of the central intelligent scheduling controller 11. In order to achieve the rapid exchange of the traffic.

於一實施例中,本發明之前端緩衝器設備10可為2n*n之可程式化的L2網路交換機,且對應該前端緩衝器設備10之設計,該光流量交換機12可為n*n之核心光流量交換機。 In one embodiment, the front end buffer device 10 of the present invention may be a 2n*n programmable L2 network switch, and corresponding to the design of the front end buffer device 10, the optical flow switch 12 may be n*n The core optical flow switch.

於另一實施例中,當中央智能排程控制器11判斷流量訊務未有衝突情況時,則傳送直通指令至前端緩衝器設備10,以令該前端緩衝器設備10將該流量訊務直接傳遞至該光流量交換機12。 In another embodiment, when the central intelligent scheduling controller 11 determines that the traffic signal does not have a conflict condition, the pass-through command is transmitted to the front-end buffer device 10, so that the front-end buffer device 10 directly directs the traffic signal. Passed to the optical flow switch 12.

具體來說,前端緩衝器設備10接受SDN(Software Defined Networking)為基礎的中央智能排程機制的管控,若是輸入流量訊務沒有衝突,則中央智能排程控制器11則執行直通(pass-through)指令給前端緩衝器設備10,令輸入的該流量訊務可直接傳遞(bypass)至光流量交換機12,若是輸入的流量訊務有衝突發生,可能是輸入埠(port)數量不夠支援同時太多的輸入流量訊務,又或者是同瞬間有2個以上的流量訊務要到同一個輸出埠時,上述情況發生時,就會產出流量訊務衝突。此狀況應由中央智能排程控制器11 來管控,若是衝突發生時,中央智能排程控制器11就啟動其中的排程演算法來管控發生衝突的流量訊務。 Specifically, the front-end buffer device 10 is controlled by a central intelligent scheduling mechanism based on SDN (Software Defined Networking). If there is no conflict in the input traffic, the central intelligent scheduling controller 11 performs pass-through. The instruction is sent to the front-end buffer device 10, so that the input traffic can be directly passed to the optical flow switch 12. If the incoming traffic has a conflict, the number of input ports may not be enough. When multiple input traffic messages are used, or when there are more than two traffic messages in the same instant to the same output port, when the above situation occurs, traffic traffic conflicts will be generated. This condition should be determined by the central intelligent scheduling controller 11 To control, if a conflict occurs, the central intelligent scheduling controller 11 activates the scheduling algorithm to control the conflicting traffic traffic.

中央智能排程控制器中的智能排程演算法可針對產生衝突之流量訊務,提供該些流量訊務優先順序判斷以及發出對前端緩衝器設備10之控管指令,以控制前端緩衝器設備10之運作。 The intelligent scheduling algorithm in the central intelligent scheduling controller can provide the traffic traffic priority order judgment and issue the control command to the front-end buffer device 10 for generating conflicting traffic signals to control the front-end buffer device. 10 operation.

智能排程演算法可包括檢查產生衝突之流量訊務的優先權,以令高優先權者直接進入該光流量交換機,以及令低優先權者進入該前端緩衝器設備以進行緩衝處理。 The intelligent scheduling algorithm may include checking the priority of the conflicting traffic traffic so that the high priority person directly enters the optical traffic switch and the low priority person enters the front end buffer device for buffering.

當流量訊務之優先權相同時,則進一步檢查該流量訊務之級別長度,以令級別大之流量訊務具有較高的優先權,俾優先從該前端緩衝器設備進入該光流量交換機,而令級別小之流量訊務繼續進入該前端緩衝器設備以進行緩衝處理。 When the priority of the traffic message is the same, the level of the traffic message is further checked to make the traffic of the level of traffic have a higher priority, and the optical traffic switch is preferentially accessed from the front-end buffer device. The traffic with a small level continues to enter the front-end buffer device for buffering.

當流量訊務之級別長度相同時,則檢查時間排程,並依據時間軸的FIFO,將先進入FIFO排程之流量訊務優先處理。另外,智能排程演算法更於該流量訊務進入該FIFO排程時,每隔一段時間將該流量訊務在該FIFO排程的優先權增加。 When the level of the traffic traffic is the same, the time schedule is checked, and according to the FIFO of the time axis, the traffic traffic that first enters the FIFO schedule is prioritized. In addition, the intelligent scheduling algorithm increases the priority of the traffic message in the FIFO schedule at intervals when the traffic message enters the FIFO schedule.

有關智能排程演算法的判斷和管控,將於之後再詳述。 The judgment and control of the intelligent scheduling algorithm will be detailed later.

於一實施例中,中央智能排程控制器11係預先建立該虛擬光路徑,以透過時間交換機制操控該光流量交換機進行該虛擬光電路交換。具體來說,中央智能排程控制器11 是一個集中式具有智慧型排程之中央控制器,具備SDN能力的智慧型控制器(smart controller),於本系統中,中央智能排程控制器11分別與前端緩衝器設備10、光流量交換機12等溝通,視流量訊務衝突情況是否發生,來啟動智能排程演算法,利用智能排程演算法的4種排程機制,以管控前端緩衝器設備10之運作,之後,中央智能排程控制器11接著建立虛擬光路徑,來操控光流量交換機12進行虛擬光電路交換,藉此達到光訊務快速交換之目的。 In an embodiment, the central intelligent scheduling controller 11 pre-establishes the virtual optical path to control the optical optical switch to perform the virtual optical circuit switching through a time switch mechanism. Specifically, the central intelligent scheduling controller 11 It is a centralized smart controller with a central controller and SDN-capable smart controller. In this system, the central intelligent scheduling controller 11 and the front-end buffer device 10 and the optical flow switch respectively 12, etc., depending on whether the traffic conflict situation occurs, to start the intelligent scheduling algorithm, using the four scheduling mechanisms of the intelligent scheduling algorithm to control the operation of the front-end buffer device 10, and then, the central intelligent scheduling The controller 11 then establishes a virtual optical path to control the optical flow switch 12 to perform virtual optical circuit switching, thereby achieving the purpose of fast optical communication exchange.

於一實施例中,光流量交換機12接收該前端緩衝器設備10所傳送之流量訊務,俾依據該中央智能排程控制器11預先設定的光交換路徑,進行該流量訊務的交換與傳送。具體來說,光流量交換機12可為全光的流量交換機(Optical Flow Switch),它的控制面是接受中央智能排程控制器11而被管控,亦即由中央智能排程控制器11通知光流量交換機12如何來進行光路光訊務的交換與傳送。 In an embodiment, the optical traffic switch 12 receives the traffic information transmitted by the front-end buffer device 10, and performs the exchange and transmission of the traffic information according to the optical switching path preset by the central intelligent scheduling controller 11. . Specifically, the optical flow switch 12 can be an all-optical flow switch, and its control surface is controlled by the central intelligent scheduling controller 11, that is, the central intelligent scheduling controller 11 notifies the light. How the traffic switch 12 performs the exchange and transmission of optical optical services.

中央智能排程控制器11是預先建立虛擬光路徑(virtual light path),來操控並排程光流量交換機12進行虛擬光電路交換(virtual circuit switching),藉此達到光訊務快速交換的任務與目的,中央智能排程控制器11還利用智能排程機制,提高系統整體交換的交換總量(Throughput)。光流量交換機12前端是接上前端緩衝器設備10,當前端緩衝器設備10送出流量訊務(如光流量訊務)時,光流量交換機12就依照中央智能排程控制器11預先設定的光交換路徑,來進行該光訊務的交換與傳送。 The central intelligent scheduling controller 11 pre-establishes a virtual light path to control and schedule the optical flow switch 12 for virtual circuit switching, thereby achieving the task and purpose of fast optical communication exchange. The central intelligent scheduling controller 11 also utilizes an intelligent scheduling mechanism to increase the total amount of exchanges (Throughput) of the system. The front end of the optical flow switch 12 is connected to the front end buffer device 10. When the current end buffer device 10 sends out traffic traffic (such as optical traffic), the optical flow switch 12 follows the preset light of the central intelligent scheduling controller 11. The path is exchanged for the exchange and transmission of the optical information.

由上可知,本發明為一種運用於資料中心網路上之光交換系統,其主要利用可程式化光流量交換機應用在資料中心上,利用智慧型排程機制技術加上可程式化光流量交換機的系統設計,使得可程式化核心光流量交換機之交換埠數量,可較原來光交換機之交換埠數量(如256)降為一半的交換埠數量(如128),但整體光流量交換機之交換總量較原光交換機從原來50%提昇到60%,另外,若交換埠數量降為3/4的交換埠數量(如192)時,交換總量則從原來50%更進一步提昇到99%。 As can be seen from the above, the present invention is an optical switching system applied to a data center network, which is mainly applied to a data center by using a programmable optical flow switch, and utilizes a smart scheduling mechanism technology and a programmable optical flow switch. The system design allows the number of exchangeable ports of the programmable optical optical switch to be reduced to half the number of switching ports (such as 128) compared to the original optical switch (such as 256), but the total amount of switching of the overall optical traffic switch Compared with the original optical switch from the original 50% to 60%, in addition, if the number of exchanges reduced to 3 / 4 exchanges (such as 192), the total amount of exchange is further increased from the original 50% to 99%.

請參閱第2A和2B圖,其所示為本發明使用128*128的核心光流量交換機以及傳統256*256的核心光流量交換機的架構示意圖,其中,第2A圖為傳統256*256的核心光流量交換機的架構圖,其中,傳統系統是採用2n*2n光流量交換機22搭配2n*2n網路交換機20,另外,第2B圖為本發明使用128*128的核心光流量交換機的架構圖,其中,本發明則是採用n*n光流量交換機12搭配2n*n前端緩衝器設備10。 Please refer to FIG. 2A and FIG. 2B, which are schematic diagrams showing the architecture of a 128*128 core optical traffic switch and a traditional 256*256 core optical traffic switch according to the present invention. FIG. 2A is a conventional 256*256 core light. The architecture diagram of the traffic switch, wherein the traditional system uses a 2n*2n optical traffic switch 22 with a 2n*2n network switch 20. In addition, FIG. 2B is an architectural diagram of a core optical traffic switch using 128*128 in the present invention, wherein In the present invention, the n*n optical flow switch 12 is used with the 2n*n front end buffer device 10.

請同時參考第2A和2B圖,本發明所提出的系統設計,配合中央智能排程控制器11內慧型排程機制技術,在前端緩衝器設備10(例如可程式化的L2網路交換機)上進行以SDN為基礎的智能排程機制的管控,將最適當的流量訊務經過此智慧型排程機制下,送往光流量交換機12進行光交換傳送。在本發明系統在配合智能排程機制管控下,原來需要256*256的核心2n*2n光流量交換機22,現在可 僅用128*128的核心光流量交換機12來取代,如此,可較原來光流量交換機之交換埠數量減少50%,但整體光流量交換機之交換總量,可從原來50%的交換總量,提昇到60%的交換總量,此較原光交換機增加20%的交換總量效能,如第3圖所示,另外,若交換埠數量降為3/4的交換埠數量(如192*192)時,則交換總量從原來50%更進一步提昇到99%,較傳統的2n*2n光流量交換機22增加接近1倍的交換總量效能,如第3圖所示。 Please refer to FIGS. 2A and 2B at the same time, the system design proposed by the present invention cooperates with the smart scheduling mechanism technology in the central intelligent scheduling controller 11 in the front-end buffer device 10 (for example, a programmable L2 network switch). The control of the SDN-based intelligent scheduling mechanism is performed, and the most appropriate traffic traffic is sent to the optical traffic switch 12 for optical switching transmission through the intelligent scheduling mechanism. In the system of the present invention, under the control of the intelligent scheduling mechanism, the original 2n*2n optical flow switch 22 of 256*256 is required, and now It is replaced by only 128*128 core optical traffic switch 12, so that the number of exchanges of the original optical traffic switch can be reduced by 50%, but the total exchange volume of the overall optical traffic switch can be 50% of the total exchange amount. Increase the total amount of exchange to 60%, which is 20% more than the original optical switch, as shown in Figure 3, in addition, if the number of exchanges is reduced to 3/4 of the number of exchanges (such as 192 * 192 When the total amount of exchange is further increased from the original 50% to 99%, it is nearly double the exchange total performance compared with the traditional 2n*2n optical flow switch 22, as shown in Figure 3.

由上可知,本發明提出一個包括可程式化光流量交換機的光交換系統,配合智慧型排程機制技術,可對前端緩衝器設備進行智能排程機制的管控,將最適當的流量訊務經過智慧型排程機制下,送往光流量交換機進行光交換傳送,在本發明下,可程式化光流量交換機之交換埠數量,可較原來光流量交換機之交換埠數量大幅減少,但整體光流量交換機之交換總量較原先光交換機卻顯著增加。 As can be seen from the above, the present invention proposes an optical switching system including a programmable optical flow switch, and with intelligent scheduling mechanism technology, the front-end buffer device can be controlled by an intelligent scheduling mechanism, and the most appropriate traffic signal can be passed. Under the intelligent scheduling mechanism, the optical traffic switch is sent to the optical traffic switch for optical switching transmission. Under the present invention, the number of exchangeable optical traffic switches can be significantly reduced compared with the original optical traffic switch, but the overall optical traffic is reduced. The total amount of exchanges of switches has increased significantly compared to the original optical switches.

中央智能排程控制器其上的智能排程演算法包括四種排程機制,第一個是檢查衝突訊務(flow)的優先權(Priority),優先權高的優先通過,直接進入光流量交換機,優先權低的訊務則進入前端緩衝器設備,進行緩衝處理。若是相同的優先權,則進行第二個檢查排程,即判斷訊務級別長短(Level Length),其中,訊務級別大的訊務具有較高的優先權,可優先從前端緩衝器設備進入光流量交換機,反之,訊務級別較短的訊務,則繼續進入前端緩衝器設備,進行緩衝處理。若衝突訊務屬同一個級別的訊務長 度,則是具有同樣的優先權,則要進行第三個檢查排程,即判斷時間軸的FIFO,此時就是進入FIFO排程,依照時間先到先處理。最後,第四個排程是採用時間FIFO排程時,當訊務進入FIFO排程後,每隔一段時間將其在FIFO的優先權增加,以避免有訊務餓死(starvation)的情況發生。 The intelligent scheduling algorithm on the central intelligent scheduling controller includes four scheduling mechanisms. The first one is to check the priority of the conflicting traffic, and the priority is passed first, and the optical traffic is directly entered. The switch, the low priority traffic enters the front-end buffer device for buffer processing. If the same priority is given, the second check schedule is performed, that is, the level of the traffic level is determined. The traffic with a large traffic level has a higher priority and can be preferentially accessed from the front-end buffer device. The optical traffic switch, on the other hand, the traffic with a shorter traffic level continues to enter the front-end buffer device for buffer processing. If the conflicting traffic is at the same level of the chief of affairs Degree, if it has the same priority, it is necessary to carry out the third inspection schedule, that is, to determine the timeline FIFO, at this time, it is to enter the FIFO schedule, according to the time first-come first processing. Finally, the fourth schedule is when the time FIFO schedule is used. When the traffic enters the FIFO schedule, the priority of the FIFO is increased at intervals to avoid the starvation of the traffic.

中央智能排程控制器就啟動其上的排程演算法詳細說明如下:Algorithm:Intelligent Scheduler while(true)do{/* do forever */於一段flow偵查時間區間內,接收所有輸入之flow;將所收到flow依據其priority,加入multi-level FIFO queue;於multi-level FIFO queue,找出priority最高之flow;若priority最高之flow數=1,將所找到flow交換至輸出埠輸出;若priority最高之flow數>1,由其中找出level_length最大之flow;若level_length最大之flow數=1,將所找到flow交換至輸出埠輸出;若level_length最大之flow數>1,由其中找出arriving time最早之flow;將所找到flow交換至輸出埠輸出; 其他未輸出之flow進入光緩衝器延遲輸出;定期將其他未輸出flow之priority增加;}end while The central intelligent scheduling controller starts the scheduling algorithm on it as follows: Algorithm:Intelligent Scheduler while(true)do{/* do forever */ receives all input flows in a period of flow detection time; The received flow is added to the multi-level FIFO queue according to its priority; in the multi-level FIFO queue, the highest priority flow is found; if the highest priority flow number=1, the found flow is switched to the output output; if priority The highest flow number is >1, and the flow with the largest level_length is found out; if the maximum flow number of level_length=1, the flow found is switched to the output output; if the maximum flow number of level_length is >1, the arrival time is found out from it. The earliest flow; exchange the found flow to the output 埠 output; Other unoutputted flows enter the optical buffer delay output; periodically increase the priority of other unoutput flows; }end while

請參閱第4圖,其所示為本發明之運用於資料中心網路之光交換系統中智能排程演算法的檢測流程。如圖所示,於流程S41中,即接收所有輸入之流量訊務。於流程S42中,即判斷優先權高低,當優先權最大之流量訊務的數量大於1時,則進入流程S43,反之,若僅有一個,則進入流程S45,即該流量訊務為要輸出之流量訊務。於流程S43中,即判斷級別長度,當即別長度最大之流量訊務的數量大於1時,則進入流程S44,反之,若僅有一個,則進入流程S45,即該流量訊務為要輸出之流量訊務。於流程S44中,即流量訊務到達時間,並找出流量訊務最早者,即為要輸出之流量訊務。 Please refer to FIG. 4, which shows the detection process of the intelligent scheduling algorithm in the optical switching system applied to the data center network of the present invention. As shown, in flow S41, all incoming traffic messages are received. In the process S42, the priority is determined. When the number of the highest priority traffic is greater than 1, the process proceeds to the process S43. Otherwise, if there is only one, the process proceeds to the process S45, that is, the traffic is output. Traffic traffic. In the process S43, the length of the level is determined, and when the number of the traffic services with the largest length is greater than 1, the process proceeds to the process S44. Otherwise, if there is only one, the process proceeds to the process S45, that is, the traffic message is output. Traffic traffic. In the process S44, that is, the traffic traffic arrival time, and find the earliest traffic traffic, that is, the traffic traffic to be output.

於流程S45中,該流程即找到該流量訊務並交換至輸出埠輸出,其他未輸出之流量訊務進入光緩衝器延遲輸出,並定期將其他未輸出流量訊務之優先權增加。換言之,即便無法輸出的流量訊務,也會逐漸提高其優先權,避免訊務餓死的情況發生。 In the process S45, the flow finds the traffic and exchanges it to the output, and the other unoutputted traffic enters the optical buffer delay output, and periodically increases the priority of other unoutput traffic. In other words, even if the traffic traffic that cannot be output will gradually increase its priority, it will prevent the starvation of the traffic.

本發明為運用於資料中心網路上之光交換系統,透過可程式化光流量交換機應用在資料中心上,以及搭配智慧型排程機制技術,以完成一種創新的光交換系統。本發明 與其他習用技術相互比較下,更具備下列兩項優點。 The invention is applied to an optical switching system on a data center network, and is applied to a data center through a programmable optical flow switch, and is matched with a smart scheduling mechanism technology to complete an innovative optical switching system. this invention Compared with other conventional technologies, it has the following two advantages.

第一,本系統配合中央智能排程機制管控下,原來需要2n*2n(如256*256)的核心光流量交換機,現在可僅用n*n(如128*128)的核心光流量交換機來取代,如此,可較原來光交換機之交換埠數量減少50%((2n-n)/2n * 100%),但整體光流量交換機之交換總量從原來的50%進步到60%,此較原光交換機增加20%的交換總量效能。另外,若用交換埠數量為原3/4的交換埠數量的核心光流量交換機來取代(如192*192),如此,可較原來光交換機之交換埠數量減少25%((256-192)/256 * 100%),但整體光流量交換機之交換總量從原來的50%提昇到99%,此較原光交換機增加接近1倍的交換總量效能。 First, the system is under the control of the central intelligent scheduling mechanism. The original optical flow switch that originally required 2n*2n (such as 256*256) can now use only the n*n (such as 128*128) core optical flow switch. Instead, this can reduce the number of exchanges compared to the original optical switch by 50% ((2n-n)/2n * 100%), but the total exchange volume of the overall optical flow switch has improved from the original 50% to 60%. The original optical switch increases the total switching performance by 20%. In addition, if you replace the core optical traffic switch with the number of switched ports of the original 3/4 (such as 192*192), you can reduce the number of switching ports compared to the original optical switch by 25% ((256-192) /256 * 100%), but the total exchange volume of the overall optical traffic switch has increased from the original 50% to 99%, which is nearly double the total switching performance of the original optical switch.

第二,本發明提出訊務流量排程機制,可強化光流量交換機之前端緩衝器設備,目前光流量交換機之前端緩衝器並未支援智慧型排程機制。本發明所提出之智能排程演算法,包括四種排程機制,第一個是檢查衝突訊務的優先權高低,優先權高的優先通過,直接進入光流量交換機,優先權低的訊務則進入前端緩衝器設備,進行緩衝處理。若是相同的優先權,則進行第二個檢查排程,即考慮訊務級別長短,訊務級別大的訊務具有較高的優先權,可優先從前端緩衝器設備進入光流量交換機,否則則是較短的訊務,則繼續進入前端緩衝器設備,進行緩衝處理。若屬同一個訊務級別的訊務長度,則是有同樣的優先權,則要進行第三個檢查排程,加入時間軸的FIFO,此時就是進入 FIFO排程,依照時間先到先處理。最後,第四個排程是時間FIFO排程,當訊務進入FIFO排程時,每隔一段時間將其在FIFO的優先權增加,以避免有訊務餓死情況產生。 Second, the present invention proposes a traffic scheduling mechanism that can enhance the front-end buffer device of the optical traffic switch. Currently, the front-end buffer of the optical traffic switch does not support the intelligent scheduling mechanism. The intelligent scheduling algorithm proposed by the invention includes four scheduling mechanisms. The first one is to check the priority of the conflicting traffic, the priority of the high priority is passed, and the optical traffic switch is directly accessed, and the low priority traffic is used. Then enter the front-end buffer device for buffer processing. If the same priority is given, the second check schedule is performed, that is, considering the length of the traffic level, the traffic with a large traffic level has a higher priority, and the optical traffic switch can be preferentially accessed from the front-end buffer device, otherwise If it is a short message, it will continue to enter the front-end buffer device for buffer processing. If it is the same traffic level of the traffic length, it has the same priority, then the third check schedule, join the timeline FIFO, then enter FIFO scheduling, processing first in a timely manner. Finally, the fourth schedule is the time FIFO schedule. When the traffic enters the FIFO schedule, it increases its priority in the FIFO at regular intervals to avoid starvation of the traffic.

上列詳細說明乃針對本發明之一可行實施例進行具體說明,惟該實施例並非用以限制本發明之專利範圍,凡未脫離本發明技藝精神所為之等效實施或變更,均應包含於本案之專利範圍中。 The detailed description of the present invention is intended to be illustrative of a preferred embodiment of the invention, and is not intended to limit the scope of the invention. The patent scope of this case.

Claims (9)

一種運用於資料中心網路之光交換系統,其包含:前端緩衝器設備,係接收來自一伺服器群的流量訊務;中央智能排程控制器,係連接該前端緩衝器設備,用以依據該前端緩衝器設備所接收之流量訊務,判斷該流量訊務是否具衝突情況,以於判斷出該流量訊務具衝突情況時,啟動智能排程演算法以管控該前端緩衝器設備之運作,並建立虛擬光路徑;以及光流量交換機,係連接該中央智能排程控制器與該前端緩衝器設備,用以依據該中央智能排程控制器之管控,利用該中央智能排程控制器建立之該虛擬光路徑執行虛擬光電路交換,俾達到該流量訊務之快速交換,其中,該中央智能排程控制器判斷該流量訊務未有衝突情況時,傳送直通指令至該前端緩衝器設備,以令該前端緩衝器設備將該流量訊務直接傳遞至該光流量交換機。An optical switching system for a data center network, comprising: a front-end buffer device for receiving traffic traffic from a server group; and a central intelligent scheduling controller connected to the front-end buffer device for The traffic signal received by the front-end buffer device determines whether the traffic signal has a conflict condition, so as to determine that the traffic signal has a conflict condition, start an intelligent scheduling algorithm to control the operation of the front-end buffer device. And establishing a virtual light path; and an optical flow switch connecting the central intelligent scheduling controller and the front-end buffer device for establishing, according to the central intelligent scheduling controller, using the central intelligent scheduling controller The virtual optical path performs virtual optical circuit switching to achieve fast switching of the traffic information, wherein the central intelligent scheduling controller determines that the traffic signal does not have a conflict condition, and transmits a through command to the front end buffer device. So that the front-end buffer device directly transmits the traffic signal to the optical traffic switch. 如申請專利範圍第1項所述之光交換系統,其中,該中央智能排程控制器係預先建立該虛擬光路徑,以透過時間交換機制操控該光流量交換機進行該虛擬光電路交換。The optical switching system of claim 1, wherein the central intelligent scheduling controller pre-establishes the virtual optical path to control the optical optical switch to perform the virtual optical circuit switching through a time switch mechanism. 如申請專利範圍第1項所述之光交換系統,其中,該光流量交換機接收該前端緩衝器設備所傳送之流量訊務,俾依據該中央智能排程控制器預先設定的光交換路徑,進行該流量訊務的交換與傳送。The optical switching system of claim 1, wherein the optical traffic switch receives the traffic information transmitted by the front-end buffer device, and performs the optical switching path preset according to the central intelligent scheduling controller. The exchange and transmission of the traffic. 一種運用於資料中心網路之光交換系統,其包含:前端緩衝器設備,係接收來自一伺服器群的流量訊務;中央智能排程控制器,係連接該前端緩衝器設備,用以依據該前端緩衝器設備所接收之流量訊務,判斷該流量訊務是否具衝突情況,以於判斷出該流量訊務具衝突情況時,啟動智能排程演算法以管控該前端緩衝器設備之運作,並建立虛擬光路徑;以及光流量交換機,係連接該中央智能排程控制器與該前端緩衝器設備,用以依據該中央智能排程控制器之管控,利用該中央智能排程控制器建立之該虛擬光路徑執行虛擬光電路交換,俾達到該流量訊務之快速交換,其中,該前端緩衝器設備為2n*n之可程式化的L2網路交換機。An optical switching system for a data center network, comprising: a front-end buffer device for receiving traffic traffic from a server group; and a central intelligent scheduling controller connected to the front-end buffer device for The traffic signal received by the front-end buffer device determines whether the traffic signal has a conflict condition, so as to determine that the traffic signal has a conflict condition, start an intelligent scheduling algorithm to control the operation of the front-end buffer device. And establishing a virtual light path; and an optical flow switch connecting the central intelligent scheduling controller and the front-end buffer device for establishing, according to the central intelligent scheduling controller, using the central intelligent scheduling controller The virtual optical path performs virtual optical circuit switching to achieve fast switching of the traffic, wherein the front-end buffer device is a 2n*n programmable L2 network switch. 如申請專利範圍第4項所述之光交換系統,其中,該光流量交換機為n*n之核心光流量交換機。The optical switching system of claim 4, wherein the optical traffic switch is a n*n core optical traffic switch. 一種運用於資料中心網路之光交換系統,其包含:前端緩衝器設備,係接收來自一伺服器群的流量訊務;中央智能排程控制器,係連接該前端緩衝器設備,用以依據該前端緩衝器設備所接收之流量訊務,判斷該流量訊務是否具衝突情況,以於判斷出該流量訊務具衝突情況時,啟動智能排程演算法以管控該前端緩衝器設備之運作,並建立虛擬光路徑;以及光流量交換機,係連接該中央智能排程控制器與該前端緩衝器設備,用以依據該中央智能排程控制器之管控,利用該中央智能排程控制器建立之該虛擬光路徑執行虛擬光電路交換,俾達到該流量訊務之快速交換,其中,該智能排程演算法包括檢查產生衝突之流量訊務的優先權,以令高優先權者直接進入該光流量交換機,以及令低優先權者進入該前端緩衝器設備以進行緩衝處理。An optical switching system for a data center network, comprising: a front-end buffer device for receiving traffic traffic from a server group; and a central intelligent scheduling controller connected to the front-end buffer device for The traffic signal received by the front-end buffer device determines whether the traffic signal has a conflict condition, so as to determine that the traffic signal has a conflict condition, start an intelligent scheduling algorithm to control the operation of the front-end buffer device. And establishing a virtual light path; and an optical flow switch connecting the central intelligent scheduling controller and the front-end buffer device for establishing, according to the central intelligent scheduling controller, using the central intelligent scheduling controller The virtual optical path performs virtual optical circuit switching to achieve fast switching of the traffic information, wherein the intelligent scheduling algorithm includes checking the priority of the conflicting traffic traffic, so that the high priority person directly enters the The optical traffic switch, and the low priority person enters the front end buffer device for buffer processing. 如申請專利範圍第6項所述之光交換系統,其中,該智能排程演算法更包括於該流量訊務之優先權相同時,檢查該流量訊務之級別長度,以令級別大之流量訊務具有較高的優先權,俾優先從該前端緩衝器設備進入該光流量交換機,而令級別小之流量訊務繼續進入該前端緩衝器設備以進行緩衝處理。The optical switching system of claim 6, wherein the intelligent scheduling algorithm further includes checking the level length of the traffic signal when the priority of the traffic message is the same, so that the traffic of the level is large. The traffic has a higher priority, and the traffic from the front-end buffer device is preferentially entered into the optical traffic switch, so that the traffic with a small level continues to enter the front-end buffer device for buffer processing. 如申請專利範圍第7項所述之光交換系統,其中,該智能排程演算法更包括於該流量訊務之級別長度相同時,檢查時間排程,以依據時間軸的FIFO,將先進入FIFO排程之流量訊務優先處理。The optical switching system of claim 7, wherein the intelligent scheduling algorithm further includes checking the time schedule when the level of the traffic signal is the same, so that the FIFO according to the time axis will enter first. The traffic of the FIFO schedule is prioritized. 如申請專利範圍第8項所述之光交換系統,其中,該智能排程演算法更包括於該流量訊務進入該FIFO排程時,每隔一段時間將該流量訊務在該FIFO排程的優先權增加。The optical switching system of claim 8, wherein the intelligent scheduling algorithm further comprises: when the traffic message enters the FIFO schedule, the traffic message is scheduled in the FIFO at intervals The priority has increased.
TW106135329A 2017-10-16 2017-10-16 System of optical switch used in data center network TWI641268B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
TW106135329A TWI641268B (en) 2017-10-16 2017-10-16 System of optical switch used in data center network

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
TW106135329A TWI641268B (en) 2017-10-16 2017-10-16 System of optical switch used in data center network

Publications (2)

Publication Number Publication Date
TWI641268B true TWI641268B (en) 2018-11-11
TW201918078A TW201918078A (en) 2019-05-01

Family

ID=65034589

Family Applications (1)

Application Number Title Priority Date Filing Date
TW106135329A TWI641268B (en) 2017-10-16 2017-10-16 System of optical switch used in data center network

Country Status (1)

Country Link
TW (1) TWI641268B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
TWI715441B (en) * 2020-02-14 2021-01-01 瑞昱半導體股份有限公司 Bandwidth allocation apparatus for use in oversubscription network switching and related network switch

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6757282B1 (en) * 1999-11-09 2004-06-29 Synchrodyne Networks, Inc. Fast switching of data packet with common time reference
US20050041970A1 (en) * 2003-07-09 2005-02-24 Nat'l Inst Of Info & Comm Tech Inc Admin Agency Optical packet buffering device and buffering method thereof
US7733895B2 (en) * 2004-11-29 2010-06-08 Cisco Technology, Inc. Non-preemptive scheduling in network elements
US20120275781A1 (en) * 2009-11-12 2012-11-01 Intune Networks Limited Switch system and method for the monitoring of virtual optical paths in an optical burst switched (obs) communication network
TWI580213B (en) * 2015-07-24 2017-04-21 Chunghwa Telecom Co Ltd A packet-programmable multi-optical delay device for light buffers
TWI600288B (en) * 2016-08-30 2017-09-21 Chunghwa Telecom Co Ltd Fiber optical switches

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6757282B1 (en) * 1999-11-09 2004-06-29 Synchrodyne Networks, Inc. Fast switching of data packet with common time reference
US20050041970A1 (en) * 2003-07-09 2005-02-24 Nat'l Inst Of Info & Comm Tech Inc Admin Agency Optical packet buffering device and buffering method thereof
US7733895B2 (en) * 2004-11-29 2010-06-08 Cisco Technology, Inc. Non-preemptive scheduling in network elements
US20120275781A1 (en) * 2009-11-12 2012-11-01 Intune Networks Limited Switch system and method for the monitoring of virtual optical paths in an optical burst switched (obs) communication network
TWI580213B (en) * 2015-07-24 2017-04-21 Chunghwa Telecom Co Ltd A packet-programmable multi-optical delay device for light buffers
TWI600288B (en) * 2016-08-30 2017-09-21 Chunghwa Telecom Co Ltd Fiber optical switches

Also Published As

Publication number Publication date
TW201918078A (en) 2019-05-01

Similar Documents

Publication Publication Date Title
US10491974B2 (en) Buffer scheduling method for flow switching
Iyer et al. Analysis of the parallel packet switch architecture
AU746166B2 (en) Fair and efficient cell scheduling in input-buffered multipoint switch
US5309426A (en) High performance cascadable simplex switch
CN110113271B (en) MPI application acceleration system and method based on photoelectric hybrid switching network
WO2011100878A1 (en) Method, switching device and system for realizing flow controlling in a switching network
KR20040038028A (en) Multiple Input/Output-Queued Switch
JPH04276847A (en) Method and system for serial-link communication
CN101212818A (en) Network node and buffer device, and scheduling method
US20050053080A1 (en) Apparatus and method for maintaining packet sequencing in a parallel router
JP2525550B2 (en) Prioritization method
CN100490383C (en) A high-speed Crossbar scheduling method for supporting multipriority
US20130148512A1 (en) Distributed control plane for link aggregation
Zhang et al. Minimizing coflow completion time in optical circuit switched networks
TWI641268B (en) System of optical switch used in data center network
US20140177648A1 (en) Tagging and synchronization for fairness in noc interconnects
Zhang et al. Reco: Efficient regularization-based coflow scheduling in optical circuit switches
CN103491023B (en) Method for routing for three-dimensional torus photoelectricity hybrid network
JP4568364B2 (en) Method, apparatus, and computer program for scheduling unicast and multicast traffic in an interconnect fabric (method and apparatus for scheduling unicast traffic and multicast traffic in an interconnect fabric)
JP2017208743A (en) Relay device and frame transfer method
Harai et al. Optical fiber-delay-line buffer management in output-buffered photonic packet switch to support service differentiation
US20030147349A1 (en) Communications systems and methods utilizing a device that performs per-service queuing
JP5627474B2 (en) Optical termination system
WO2012103704A1 (en) Multicast duplication method, device and system
JP5451861B1 (en) Method and apparatus for setting priority route in optical packet switch network

Legal Events

Date Code Title Description
MM4A Annulment or lapse of patent due to non-payment of fees