KR102652395B1 - Optical networking method and apparatus performing the same - Google Patents

Abstract

An optical networking method and apparatus for performing the same are disclosed. An optical networking method according to an embodiment includes the steps of setting an optical path and allocating memory based on input data traffic, determining whether a collision occurs between the optical paths, and determining whether a collision occurs between the optical paths and the collision. It includes controlling a plurality of modules based on whether or not they occur.

Description

Optical networking method and apparatus for performing the same {OPTICAL NETWORKING METHOD AND APPARATUS PERFORMING THE SAME}

The embodiments below relate to an optical networking method and a device for performing the same.

As cloud technology becomes more common, the amount of data that needs to be processed within data centers that accommodate it is increasing exponentially. In addition, as the need for high-speed parallel processing operations gradually increases, the current processing-centric computing method is changing to a data-centric computing method.

In addition to high power consumption and time delay issues, large-capacity data centers require expensive maintenance/repair costs to accommodate the increasing amount of data. Optical switching technology and memory disaggregation technology are being researched as a way to solve these problems of existing data centers.

Memory disaggregation technology using existing optical switches is an obstacle to commercialization because it requires a complex networking structure to control traffic congestion due to the absence of an optical buffer and requires high-speed optical header processing technology.

Embodiments may provide technology for configuring an internal data center network using optical switches. Specifically, embodiments may provide data center optical networking technology for providing memory disaggregation functionality using optical switches.

An optical networking method according to an embodiment includes the steps of setting an optical path and allocating memory based on input data traffic, determining whether a collision occurs between the optical paths, and determining whether a collision occurs between the optical paths and the collision. It includes controlling a plurality of modules based on whether or not they occur.

The plurality of modules may include at least one of an operation module, a high-speed memory module, a low-speed memory module, and a special task module.

The plurality of modules each include the same number of optical transceivers, and the optical transceivers may each be connected to a corresponding optical switch.

The setting step may include extracting header information of the data traffic, setting a light path based on the header information, and allocating available memory through the light path.

The controlling step includes allocating time slots to the plurality of modules, selecting ports included in the plurality of modules, and allocating wavelengths of optical transceivers included in the plurality of modules. May include steps.

The step of allocating the timeslot may include setting scheduling information for each allocated timeslot when the conflict occurs, and allocating time slots consecutively when the conflict does not occur.

The step of allocating the timeslot may include removing the guard time when the collision does not occur.

An optical networking device according to an embodiment sets an optical path based on a rack switch and a first control signal received from the rack switch, and allocates memory that can be used through the optical path, A network manager that allocates a time slot, a plurality of modules that select a port based on a second control signal received from the network manager and allocate a wavelength of the optical transceiver, and a plurality of modules that allocate a wavelength of the optical transceiver, and It includes a plurality of optical switches that switch and output optical signals to the output port.

The plurality of modules each include the same number of optical transceivers, and the optical transceivers may each be connected to a corresponding optical switch.

The rack switch extracts header information of input data traffic, and the first control signal may be a signal including the header information.

The second control signal may be a signal including optical path setting information, memory area information, and timeslot allocation information.

The rack switch may transmit data traffic to at least one of the plurality of modules in response to a third control signal received from the network manager.

The plurality of modules may include at least one of an operation module, a high-speed large-capacity memory module, a low-speed memory module, and a special task module.

The network manager may successively allocate time slots when no conflict occurs between the light paths, and may set scheduling information for each allocated timeslot when a conflict occurs between the light paths.

The plurality of modules may eliminate the guard time when no collision occurs between the optical paths.

The operation module updates the destination address (DA) table for each memory area based on the second control signal, and the high-speed memory module updates the source address (SA) table for each input port based on the second control signal. It can be updated.

A data center according to an embodiment includes a gateway, a spine switch, and a plurality of racks, wherein the plurality of racks includes a rack switch and a first control signal received from the rack switch. A network manager that sets an optical path, allocates memory available through the optical path, and allocates a time slot, selects a port based on a second control signal received from the network manager, and wavelength of the optical transceiver. It includes a plurality of modules that allocate and a plurality of optical switches that switch and output optical signals to different output ports according to the wavelength of the input optical signal.

1 is a schematic block diagram of a data center according to one embodiment.
FIG. 2 is a diagram for explaining in detail the operation of the optical switch shown in FIG. 1.
FIG. 3 is a schematic block diagram of the network manager shown in FIG. 1.
FIG. 4 is a diagram for explaining the connection between blocks inside the rack shown in FIG. 1.
FIG. 5 is a flowchart illustrating a process for transmitting data traffic between modules included in the data center shown in FIG. 1.
Figure 6 is a diagram showing synchronization clocks and frame pulses for generating synchronized timeslots for timeslot-based congestion control.
FIGS. 7A and 7B are diagrams for explaining an example of an operation of allocating an output port and its destination for congestion control.

Hereinafter, embodiments will be described in detail with reference to the attached drawings. However, various changes may be made to the embodiments, so the scope of the patent application is not limited or limited by these embodiments. It should be understood that all changes, equivalents, or substitutes for the embodiments are included in the scope of rights.

The terms used in the examples are for descriptive purposes only and should not be construed as limiting. Singular expressions include plural expressions unless the context clearly dictates otherwise. In this specification, terms such as “comprise” or “have” are intended to designate the presence of features, numbers, steps, operations, components, parts, or combinations thereof described in the specification, but are not intended to indicate the presence of one or more other features. It should be understood that this does not exclude in advance the possibility of the existence or addition of elements, numbers, steps, operations, components, parts, or combinations thereof.

Terms such as first or second may be used to describe various components, but the components should not be limited by the terms. The terms are used solely for the purpose of distinguishing one component from another, for example, without departing from the scope of rights according to the concept of the embodiment, a first component may be named a second component, and similarly The second component may also be referred to as the first component.

Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as generally understood by a person of ordinary skill in the technical field to which the embodiments belong. Terms defined in commonly used dictionaries should be interpreted as having a meaning consistent with the meaning in the context of the related technology, and unless explicitly defined in the present application, should not be interpreted in an ideal or excessively formal sense. No.

In addition, when describing with reference to the accompanying drawings, identical components will be assigned the same reference numerals regardless of the reference numerals, and duplicate descriptions thereof will be omitted. In describing the embodiments, if it is determined that detailed descriptions of related known technologies may unnecessarily obscure the gist of the embodiments, the detailed descriptions are omitted.

A module in this specification may mean hardware that can perform functions and operations according to each name described in this specification, or it may mean computer program code that can perform specific functions and operations. , or it may mean an electronic recording medium loaded with computer program code that can perform specific functions and operations, for example, a processor or microprocessor.

In other words, a module may mean a functional and/or structural combination of hardware for carrying out the technical idea of the present invention and/or software for driving the hardware.

1 is a schematic block diagram of a data center according to one embodiment.

The data center 10 uses only a plurality of optical switches 200 and the same number of tunable optical transceivers to effectively avoid collisions between nodes without a complex optical header processor or optical buffer to control the optical switches 200. there is.

The data center 10 can solve the explosively increasing bandwidth requirements and low-latency problems and maximize the efficiency of network resources by configuring a data center network using the optical switch 200 with low-latency and high-capacity characteristics. .

Data center 10 includes a gateway 110, a spine switch (120), and a plurality of racks (130-1, 130-2, and 130-3). Hereinafter, the description of the rack 130 applies to each of the plurality of racks 130-1, 130-2, and 130-3.

The data center 10 can be interconnected with an external network through the gateway 110, and the spine switch 120 can connect a plurality of racks 130-1, 130-2, and 130-3.

The racks 130-1, 130-2, and 130-3 constituting the internal network of the data center 10 include a ToR switch 210, an optical switch 200, a network manager 300, and one or more dCOMP ( disaggregated Computing (disaggregated computing) modules 230-1, 230-2, and 230-3, one or more disaggregated memory (dMEM) modules 250-1, one or more disaggregated storage (dStor) modules 230-2, and one or more It may include a disaggregated Accelerator (dAccel) module 250-3.

In Figure 1, each of the racks (130-1, 130-2, and 130-3) is shown as having the same type of modules and the same connection structure; however, each of the racks (130-1, 130-2, and 130-3) Each module includes different types of modules, and each included module may be arranged in a different connection structure.

The dCOMP modules 230-1, 230-2, and 230-3 are modules that perform various computational tasks, and the dMEM module 250-1 is a plurality of dCOMP modules 230-1, 230-2, and 230. -3) is a high-speed memory pool shared and used by the dStor module, and the dStor module is a low-speed memory pool shared and used by multiple dCOMP modules (230-1, 230-2, and 230-3), and the dAccel module is a It may be a module optimized for performing a specific task.

The ToR switch 210 provides a switching function between a plurality of dCOMP modules (230-1, 230-2, and 230-3), and a plurality of dCOMP modules (230-1, 230-2, and 230-3). 3) can be connected to the spine switch 120.

The optical switch 200 combines a plurality of dCOMP modules 230-1, 230-2, and 230-3 with one or more dMEM modules 250-1, one or more dStor modules 250-2, and/or one It can be connected to the above dAccel modules (250-3).

The dCOMP modules 230-1, 230-2, and 230-3 may be CPU devices that perform various computational tasks. The dCOMP modules 230-1, 230-2, and 230-3 may use internal memory to perform computational tasks and, if necessary, may use the memory of the dMEM module 250-1.

The dMEM module 250-1 may be a high-capacity memory device comprised of a plurality of memories. The dMEM module 250-1 may provide memory access functions (read/write) upon request from the dCOMP modules 230-1, 230-2, and 230-3.

FIG. 2 is a diagram for explaining in detail the operation of the optical switch shown in FIG. 1.

The optical switch 200 can switch and output optical signals to different output ports depending on the wavelength of the input optical signal.

The optical switch 200 has the characteristic that the wavelength being switched is cycled according to the input port. Therefore, the optical switch 200 can prevent the same wavelength signals from being output through the same port even when wavelength division multiplexed optical signals composed of the same wavelengths are input to different input ports, thereby preventing interference between optical signals of the same wavelength. Switching can be done without (or crashing).

FIG. 3 is a schematic block diagram of the network manager shown in FIG. 1, and FIG. 4 is a diagram for explaining the connections between blocks inside the rack shown in FIG. 1.

The network manager 300 may include a controller 400 and a synchronizer 500.

The rack 130 includes a plurality of dCOMP modules (230-1, 230-2, and 230-3), one or more dMEM/dStor/dAccel modules (250-1, ... 250-N), and a plurality of optical switches ( 200-1, 200-2, and 200-3), a network manager 300, and a ToR switch 210.

In FIG. 4, only the dMEM modules 250-1 and 250-N are shown for convenience of explanation, but the rack 130 may include a plurality of dStor modules and a plurality of dAccel modules, and the dMEM module is located at the dStor module or dAccel module can be located.

The dCOMP modules 230-1, ...230-N may include a CPU 410 and a plurality of optical transceivers 420-1, 420-2, and 420-3. In Figure 4, for convenience of explanation, only three optical transceivers (420-1, 420-2, and 420-1) are shown, but each dCOMP module (230-1, ...230-N) is the same as each other, not three. It may include any number of optical transceivers.

The dMEM modules (250-1, ...250-N) include a controller 440, a plurality of optical transceivers (430-1, 430-2, and 430-3), and a plurality of memory controllers (MC; 450-1, 450- 2, and 450-3), and a plurality of memories 460-1, 460-2, and 460-3.

In Figure 4, for convenience of explanation, three optical transceivers (430-1, 430-2, and 430-3) and memory controllers (MC; 450-1, 450-2, and 450-3) are shown, respectively. Each dMEM module (250-1, ...250-N) may include an arbitrary number of optical transceivers and memory controllers that are the same as each other, rather than three.

In FIG. 4, for convenience of explanation, each dMEM module (250-1, ...250-N) is shown as including three physically separated memories (460-1, 460-2, and 460-3). , the memories 460-1, 460-2, and 460-3 are abstractly divided memories corresponding to locations corresponding to the memory controller (MC; 450-1, 450-2, and 450-3), respectively. It could be. For example, memories 460-1, 460-2, and 460-3 may be DDR4.

Each optical transceiver (420-1, 420-2, 420-3, 430-1, 430) included in the dCOMP module (230-1, …230-N) and dMEM module (250-1, …250-N) -2, and 430-3) may be connected to the respective optical switches 200-1, 200-2, and 200-3. At this time, each optical transceiver (420-1, 420-2, 420-3, 430-1, 430-2, and 430-3) corresponds to each optical switch (200-1, 200-2, and 200). -3) It may be physically distant from the other.

The number of optical switches (200-1, 200-2, and 200-3) is determined by the optical transceivers ( 420-1, 420-2, 420-3, 430-1, 430-2, and 430-3) (for convenience of explanation, three of each are shown in FIG. 4).

The ToR switch 210 can exchange control signals with the network manager 300 and transmit data traffic to the dCOMP modules 230-1, ...230-N.

The network manager 300 may include a controller 400 and a synchronizer 500.

The controller 400 provides an optical path between the dCOMP modules 230-1, ...230-N and the dMEM modules 250-1, ...250-N in response to the first control signal received from the ToR switch 210. can be set.

The controller 400 determines the location and size of the memories 460-1, 460-2, and 460-3 in the dMEM modules 250-1, ...250-N that can be used (or accessed) through each optical path. Can be assigned.

The controller 400 may generate a second control signal and transmit it to the dCOMP modules 230-1, ...230-N and the dMEM modules 250-1, ...250-N. For example, the controller 400 transmits the second control signal to the dCOMP module (230-1, ...230-N) and the dMEM module (250-1, ...250-N) to transmit optical path setting information and memory allocation information. You can. The second control signal may also include timeslot information.

The controller 400 may transmit a third control signal to the optical switch 200 to control the optical switch 200 . The optical switch 200 may switch the input optical signal based on the third control signal transmitted from the controller 400.

The ToR switch 210 may transmit data traffic to the dCOMP modules 230-1, ...230-N after optical path setting and memory allocation are completed. For example, the ToR switch 210 may transmit data traffic to the dCOMP modules 230-1, ... 230-N after receiving a setup completion signal from the controller 400.

The dCOMP modules 230-1, ...230-N may set the optical path based on the second control signal received from the controller 400. For example, the dCOMP modules (230-1, ...230-N) select optical switches for connection with the dMEM modules (250-1, ...250-N) and set the optical path by changing the wavelength of the optical transceiver. You can.

The dCOMP module (230-1, …230-N) is one of a plurality of optical transceivers (420-1, 420-2, 420-3) and a plurality of optical switches (200-1, 200-2, and 200-3). Data traffic flowing in from other optical paths by connecting optical paths using available optical transceivers (420-1, 420-2, 420-3) and optical switches (200-1, 200-2, and 200-3) Collisions can be avoided.

The dCOMP modules 230-1, ...230-N can allocate a memory area to be accessed (read/write) using the first control signal received from the controller 400.

The dCOMP modules 230-1, ...230-N can set optical condensation based on the memory area allocation result. For example, the dCOMP modules (230-1, …230-N) can update the destination address (DA) for each input/output port and change the wavelength of the optical transceivers (420-1, 420-2, 420-3). there is.

The dCOMP module (230-1, ...230-N) can transmit a setting completion signal to the controller 400 after completing the optical path setting.

The optical switches 200-1, 200-2, and 200-3 may switch optical signals based on the wavelength of the input optical signal. Additionally, the optical switches 200-1, 200-2, and 200-3 may switch optical signals based on the third control signal received from the controller 400.

Each dMEM module (250-1, ...250-N) is configured to prevent memory areas between the plurality of dMEM modules (250-1, ...250-N) from overlapping based on the second control signal received from the controller 400. An offset value can be assigned for (250-1, …250-N).

The dMEM modules 250-1, ...250-N may transmit each allocated offset information to the controller 400.

The controller 400 may transmit offset information of the dMEM modules 250-1, ...250-N to the dCOMP modules 230-1, ...230-N. The dCOMP module (230-1, ...230-N) can use the offset information of the dMEM module (250-1, ...250-N) to set the optical path.

The dCOMP modules 230-1, ...230-N may have table values representing all memory areas included in the data center 10.

The dCOMP module (230-1, …230-N) divides the entire memory area into each dMEM module (250-1, …250-N), and multiple MCs located within the dMEM module (250-1, …250-N). Offset information can be used to separate by (Memory Controller).

That is, the dCOMP modules 230-1, ...230-N can use the offset information to find the location of the memory 460-1, 460-2, and 460-3 to which it will access, and based on this, the optical path can be set.

The controller 440 included in the dMEM module (250-1, ...250-N) switches the optical signal input from the dCOMP module (230-1, ...230-N) to the corresponding memory based on the memory offset value. .

Optical transceiver #1 (420-1) included in the dCOMP #1 module (230-1) is connected to optical switch #1 (200-1), and optical transceiver #2 (420-2) is connected to optical switch #2 (200-1). -2), and optical transceiver #3 (420-3) can be connected to optical switch #3 (200-3).

Optical transceiver #1 (430-1) included in dMEM #1 module (250-1) is connected to optical switch #1 (200-1), and optical transceiver #2 (430-2) is connected to optical switch #2 (200-1). -2), and optical transceiver #3 (430-3) can be connected to optical switch #3 (200-3).

The optical transceivers 420-1, 420-2, 420-3, 430-1, 430-2, and 430-3 are used to set an optical path based on the second control signal received from the controller 400. The wavelength of light can be changed.

That is, the dCOMP module (230-1, ...230-N) is a specific dMEM module (250-1, ...250-N) among a number of physically distant dMEM modules (250-1, ...250-N). The optical wavelength of optical transceivers 420-1, 420-2, 420-3, 430-1, 430-2, and 430-3 to use memory locations 460-1, 460-2, and 460-3. This may change.

The dCOMP#1 module (230-1) is connected to the dMEM#1 module (250-1) when the optical wavelength of optical transceiver #1 (420-1) is λ1, and when the optical wavelength is λ2, the dMEM#N (250-1) N) Can be connected to the module.

The dCOMP#N module (230-N) is connected to the dMEM#1 module (250-1) when the optical wavelength of optical transceiver #1 (421-1) is λ3, and when the optical wavelength is λ4, the dMEM#N (250-1) N) Can be connected to the module.

The dCOMP#1 module (230-1) can be connected to the dMEM#1 module (250-1) through optical transceiver #2 (420-2) and optical switch #2 (200-2), and optical transceiver #3 ( It can also be connected to the dMEM#1 module (250-1) through (420-3) and optical switch #3 (200-3).

The data center 10 has a plurality of dCOMP modules (230-N) because each dCOMP module (230-1, ...230-N) can be connected to the same dMEM module (250-1, ...250-N) through various optical paths. -1, ...230-N) are connected to the same dMEM module (250-1, ...250-N) through different optical paths at the same time, thereby avoiding collisions between data traffic.

FIG. 5 is a flowchart illustrating a process for transmitting data traffic between modules included in the data center shown in FIG. 1.

The ToR switch 210 may extract header information of the input data traffic and transmit the header information to the network manager 300 (510). Header information may include SA (source address), DA (destination address), Application, etc.

The network manager 300 may set a light path based on header information and allocate available memory through the light path (610).

The network manager 300 may determine whether congestion (collision) occurs between set optical paths (620).

If there is no congestion (collision) between optical paths, the network manager 300 continuously allocates timeslots and sends timeslot information and memory area information to the dCOMP modules 230-1, ...230-N and the dMEM module 250. -1, …250-N) can be transmitted (630).

The dCOMP modules 230-1, ...230-N may allocate a time slot to be transmitted through the optical path based on control information received from the network manager 300 (710). At this time, the dCOMP module (230-1, ...230-N) can remove the guard time. That is, the timeslot may not have a guard time.

The dCOMP module (230-1, ...230-N) can allocate a memory area to be accessed through an optical path and update the DA table for each memory area (710). Additionally, the dCOMP modules (230-1, ...230-N) can select ports and allocate wavelengths to the optical transceivers (420-1, 420-2, 420-3) (710).

When step 710 is completed, the dCOMP modules 230-1,...230-N may transmit a setup completion signal to the network manager 300 (711).

The dMEM modules 250-1, ...250-N can allocate a time slot to be transmitted through the optical path based on control information received from the network manager 300. dMEM modules (250-1, …250-N) can eliminate guard time. That is, the timeslot may not have a guard time.

The dMEM module (250-1, …250-N) can update the SA table for each input port. Additionally, the dMEM modules 250-1, ...250-N can select ports and allocate wavelengths of the optical transceivers 430-1, 430-2, and 430-3.

The dMEM modules (250-1, …250-N) include a plurality of memories (460-1, 460-2, and 460-3) to maximize memory capacity, so a plurality of MCs (450-1, 450- 2, and 450-3) can be used.

The dMEM modules (250-1, …250-N) store memories (460-1, 460-2, and 460-3) connected to each MC (450-1, 450-2, and 450-3). Offsets can be used to differentiate.

Although omitted in FIG. 5, the offset may be transmitted to the network manager 300 through a control channel during system initialization.

When step 810 is completed, the dMEM modules 250-1,...250-N may transmit a setup completion signal to the network manager 300 (811).

When the network manager 300 receives a setting completion signal from the dCOMP module (230-1, ...230-N) and the dMEM module (250-1, ...250-N), it sends the optical path setting information and the setting completion signal to the ToR switch ( 210).

When the ToR switch 210 receives optical path setting information and a setting completion signal from the network manager 300, the ToR switch 210 can transmit data traffic to the corresponding dCOMP module (230-1, ...230-N). There is (520).

When data traffic is input, the dCOMP module (230-1, ...230-N) can generate header information for each data traffic and then insert it into the data traffic and output it.

The optical switch 200 does not require a separate switch control signal and can switch optical signals to different output ports depending on the wavelength of the input optical signal. Therefore, in process 710, the wavelengths of the optical transceivers 420-1, 420-2, 420-3, 430-1, 430-2, and 430-3 can be assigned and the output port of the optical switch can be determined at the same time. .

The dMEM modules 250-1, ...250-N may extract memory information by reading header information of the input data traffic (820).

The dMEM module (250-1, ...250-N) processes the offset from the extracted memory information and converts it to a memory address in the dMEM module (250-1, ...250-N) to address the corresponding memory (460-1, 460- 2, and 460-3) can be accessed (820).

In the case of a read operation, the dMEM module (250-1, ...250-N) may determine an output port to transmit the read data to the dCOMP module (230-1, ...230-N) and then transmit data traffic.

When congestion (collision) occurs between light paths, the network manager 300 allocates a timeslot and sends the timeslot information and memory area information to the dCOMP modules 230-1, ...230-N and the dMEM modules 250-1, …250-N) can be transmitted (640).

Additionally, the network manager 300 may transmit scheduling information for each timeslot to the dCOMP modules 230-1, ...230-N and the dMEM modules 250-1, ...250-N (640).

The dCOMP modules (230-1, ...230-N) select an output port for each timeslot based on the scheduling information received from the network manager (300) and the optical transceivers (420-1, 420-2, and 420-3) The wavelength can be changed (730).

The dCOMP modules 230-1, ...230-N can allocate a time slot and guard time to be transmitted through the optical path based on control information received from the network manager 300 (730).

The dCOMP module (230-1, ...230-N) can allocate a memory area to be accessed through an optical path and update the DA table for each memory area (730).

Additionally, the dCOMP modules 230-1, ...230-N can select ports and allocate wavelengths of the optical transceivers 420-1, 420-2, and 420-3. For example, the dCOMP modules 230-1, ...230-N may allocate wavelengths to the optical transceivers 420-1, 420-2, and 420-3 for each time slot (730).

After step 730 is completed, the dCOMP modules 230-1, ...230-N may transmit a configuration completion signal to the network manager 300 (731).

The dMEM modules 250-1, ...250-N can allocate a time slot and guard time to be transmitted through the optical path based on control information received from the network manager 300 (830).

The dMEM module (250-1, …250-N) can update the SA table for each input port (830). Additionally, the dMEM modules 250-1, ...250-N may select ports and allocate wavelengths to the optical transceivers 430-1, 430-2, and 430-3 (830). The dMEM modules 250-1, ...250-N can allocate wavelengths to the optical transceivers 430-1, 430-2, and 430-3 for each time slot (830).

After step 830 is completed, the dMEM modules 250-1, ...250-N may transmit a configuration completion signal to the network manager 300 (831).

When the network manager 300 receives a setting completion signal from the dCOMP module (230-1, ...230-N) and the dMEM module (250-1, ...250-N), it sends the optical path setting information and the setting completion signal to the ToR switch ( 210).

When the ToR switch 210 receives optical path setting information and a setting completion signal from the network manager 300, the ToR switch 210 can transmit data traffic to the corresponding dCOMP module (230-1, ...230-N). there is.

When data traffic is input, the dCOMP module (230-1, ...230-N) can generate header information for each data traffic and then insert it into the data traffic and output it.

The dMEM modules 250-1, ...250-N may extract memory information by reading header information of the input data traffic (840).

The dMEM module (250-1, ...250-N) processes the offset from the extracted memory information and converts it to a memory address in the dMEM module (250-1, ...250-N) to address the corresponding memory (460-1, 460- 2, and 460-3) can be accessed (840).

In the case of a read operation, the dMEM module (250-1, ...250-N) may determine an output port to transmit the read data to the dCOMP module (230-1, ...230-N) and then transmit data traffic.

Figure 6 is a diagram showing synchronization clocks and frame pulses for generating synchronized timeslots for timeslot-based congestion control.

The synchronizer 500 can generate synchronization clocks and frame pulses.

Synchronizer 500 may be included in network manager 300, but may be implemented as a separate functional block within data center 10.

The dCOMP modules (230-1, ...230-N) and dMEM modules (250-1, ...250-N) can synchronize clocks between modules to perform congestion control based on time slots.

The frame pulse can be generated once for every certain number (N) of synchronization clocks and can be used to find the starting point of the synchronization clock.

Each functional block included in the data center 10 can recognize the moment when the frame pulse is input as the starting point of the synchronization clock.

FIGS. 7A and 7B are diagrams for explaining an example of an operation of allocating an output port and its destination for congestion control.

The horizontal axis of FIGS. 7A and 7B represents time, and the vertical axis represents output ports of the dCOMP modules 230-1, ...230-N. Each square represents a timeslot, and the size of the square can be determined as an integer multiple of unit time. Timeslots can be distinguished by guard times between timeslots, and numbers within the timeslot indicate the destination of the timeslot.

Referring to Figure 7a, output port number 1 of the dCOMP#1 module and output port number 1 of the dCOMP#2 module have the same destination (destination number 2) in the same timeslot. In this case, a collision may occur at the input of the optical transponder of the dMEM module and the received data may be damaged.

Referring to FIG. 7b, in order to prevent data damage, the data center 10 changes the first timeslot in which a collision occurs to output port 2 of the dCOMP#2 module and outputs the data. At this time, it is assumed that the priority of the output port of the dCOMP#1 module is higher than the priority of the output port of the dCOMP#2 module.

Referring to Figure 7a, output port number 1 and output port number 2 of the dCOMP#1 module and dCOMP#2 module collide at destinations 5 and 4, respectively. In the case of destination number 5, the first output port of the dCOMP#2 module is empty, so the dCOMP#2 module changes the output port to number 1 and outputs.

In the case of destination 4, since all other output ports of the dCOMP#2 module are in use, output can be performed after waiting until output port 1 is available.

The data center 10 can connect a plurality of dCOMP modules and a plurality of dMEM modules using only a plurality of optical switches and optical transceivers without optical header processing or optical buffers that require high-speed signal processing. Accordingly, the data center 10 can effectively control conflicts between data traffic that occur.

The data center 10 uses only a plurality of optical switches 200 and the same number of wavelength-tunable optical transceivers to effectively avoid collisions between nodes without an optical header processor or optical buffer to control the optical switches 200. .

The method according to the embodiment may be implemented in the form of program instructions that can be executed through various computer means and recorded on a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, etc., singly or in combination. Program instructions recorded on the medium may be specially designed and configured for the embodiment or may be known and available to those skilled in the art of computer software. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical media such as CD-ROMs and DVDs, and magnetic media such as floptical disks. -Includes optical media (magneto-optical media) and hardware devices specifically configured to store and execute program instructions, such as ROM, RAM, flash memory, etc. Examples of program instructions include machine language code, such as that produced by a compiler, as well as high-level language code that can be executed by a computer using an interpreter, etc. The hardware devices described above may be configured to operate as one or more software modules to perform the operations of the embodiments, and vice versa.

Software may include a computer program, code, instructions, or a combination of one or more of these, which may configure a processing unit to operate as desired, or may be processed independently or collectively. You can command the device. Software and/or data may be used on any type of machine, component, physical device, virtual equipment, computer storage medium or device to be interpreted by or to provide instructions or data to a processing device. , or may be permanently or temporarily embodied in a transmitted signal wave. Software may be distributed over networked computer systems and thus stored or executed in a distributed manner. Software and data may be stored on one or more computer-readable recording media.

Although the embodiments have been described with limited drawings as described above, those skilled in the art can apply various technical modifications and variations based on the above. For example, the described techniques are performed in a different order than the described method, and/or components of the described system, structure, device, circuit, etc. are combined or combined in a different form than the described method, or other components are used. Alternatively, appropriate results may be achieved even if substituted or substituted by an equivalent.

Therefore, other implementations, other embodiments, and equivalents of the claims also fall within the scope of the following claims.

Claims (17)

delete delete delete delete delete delete delete rack switch;
a network manager that sets an optical path based on a first control signal received from the rack switch, allocates memory that can be used through the optical path, and allocates a time slot;
a plurality of modules that select a port and allocate a wavelength of an optical transceiver based on a second control signal received from the network manager; and
A plurality of optical switches that switch and output optical signals to different output ports depending on the wavelength of the optical signal being input.
Optical networking device including.
According to clause 8,
The plurality of modules are,
Each includes the same number of optical transceivers,
The optical transceivers are,
Optical networking devices each connected to a corresponding optical switch.
According to clause 8,
The rack switch is,
Extract header information from input data traffic,
The first control signal is,
An optical networking device that is a signal including the header information.
According to clause 8,
The second control signal is,
An optical networking device whose signal includes optical path setting information, memory area information, and timeslot allocation information.
According to clause 8,
The rack switch is,
An optical networking device that transmits data traffic to at least one of the plurality of modules in response to a third control signal received from the network manager.
According to clause 8,
The plurality of modules are,
An optical networking device including at least one of an operation module, a high-speed large-capacity memory module, a low-speed memory module, and a special task module.
According to clause 8,
The network manager is,
An optical networking device that allocates time slots consecutively when no collision occurs between the optical paths, and sets scheduling information for each allocated timeslot when a collision occurs between the optical paths.
According to clause 8,
The plurality of modules are,
An optical networking device that eliminates guard time when no collision occurs between the optical paths.
In paragraph 13
The calculation module is,
Update the destination address (DA) table for each memory area based on the second control signal,
The high-speed memory module,
Updating the source address (SA) table for each input port based on the second control signal.
Optical networking devices.
delete