WO2021249422A1 - 服务器 - Google Patents

服务器 Download PDF

Info

Publication number
WO2021249422A1
WO2021249422A1 PCT/CN2021/099082 CN2021099082W WO2021249422A1 WO 2021249422 A1 WO2021249422 A1 WO 2021249422A1 CN 2021099082 W CN2021099082 W CN 2021099082W WO 2021249422 A1 WO2021249422 A1 WO 2021249422A1
Authority
WO
WIPO (PCT)
Prior art keywords
unit
insertion mechanism
server
connector
blade
Prior art date
Application number
PCT/CN2021/099082
Other languages
English (en)
French (fr)
Inventor
赵闪
田海东
熊先奎
Original Assignee
中兴通讯股份有限公司
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 中兴通讯股份有限公司 filed Critical 中兴通讯股份有限公司
Publication of WO2021249422A1 publication Critical patent/WO2021249422A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F1/00Details not covered by groups G06F3/00 - G06F13/00 and G06F21/00
    • G06F1/16Constructional details or arrangements
    • G06F1/18Packaging or power distribution
    • G06F1/183Internal mounting support structures, e.g. for printed circuit boards, internal connecting means
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F1/00Details not covered by groups G06F3/00 - G06F13/00 and G06F21/00
    • G06F1/16Constructional details or arrangements
    • G06F1/18Packaging or power distribution
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L12/00Data switching networks
    • H04L12/02Details
    • H04L12/04Switchboards

Definitions

  • the present disclosure relates to the field of computer technology, and in particular to a server.
  • Blade servers are widely used in the fields of telecommunications, supercomputers, etc., with the characteristics of high integration, high processing capacity, shared heat dissipation and power supply.
  • the blade unit and the switching unit are connected through the backplane. Therefore, the signal rate and signal bandwidth are largely dependent on the backplane.
  • Blade servers are widely used in the fields of telecommunications, supercomputers, etc., with the characteristics of high integration, high processing capacity, shared heat dissipation and power supply.
  • the blade unit and the switching unit are connected through the backplane. Therefore, the signal rate and signal bandwidth are largely dependent on the backplane.
  • the embodiment of the present disclosure proposes a server, which aims to increase the bandwidth and transmission rate of the server, and increase the flexibility of the internal interface.
  • the server provided by the embodiment of the present disclosure includes a front insertion mechanism and a rear insertion mechanism, the front insertion mechanism includes a blade unit, the rear insertion mechanism includes a main exchange unit, and the blade unit and the main exchange unit are connected to each other through a connector Orthogonal connection.
  • Figure 1 is a side view of a server provided by an embodiment of the disclosure
  • Figure 2 is a front view of a front insertion mechanism provided by an embodiment of the disclosure
  • Figure 3 is a rear view of the rear insertion mechanism provided by an embodiment of the disclosure.
  • FIG. 4 is a schematic diagram of a part of the structure of a liquid cooling module provided by the implementation of the present disclosure
  • Figure 5 is a rear view of a rear insertion mechanism provided by another embodiment of the present disclosure.
  • Fig. 6 is a side view of a server provided by another embodiment of the present disclosure.
  • 10-blade unit 20-main exchange unit, 30-connector, 40-cooling unit, 41-air cooling module, 42-liquid cooling module, 421-water separator, 422-water tray, 423-water inlet pipe, 424-outlet pipe, 50-power supply, 60-frame management unit, 70-auxiliary exchange unit, 80-backplane.
  • the server includes multiple blade units and switching units.
  • the blade units and switching units implement high-speed signal transmission through a backplane.
  • blade units need to be integrated.
  • the bandwidth of the interface between the blade unit and the switching unit is limited, and after the backplane is designed, it cannot be expanded.
  • the embodiment of the present disclosure provides a server, which can not only integrate blade units together, but also increase the interface bandwidth between the blade unit and the switching unit, and the server can be expanded.
  • FIG. 1 is a side view of a server provided by an embodiment of the disclosure
  • FIG. 2 is a front view of the front insertion mechanism of the server in FIG. 1.
  • the server includes a front insertion mechanism and a rear insertion mechanism, wherein the front insertion mechanism includes a blade unit 10, and the rear insertion mechanism includes a main exchange unit 20, and the blade unit 10 and the main exchange unit 20 are orthogonal to each other through a connector 30 Way to connect.
  • the blade unit 10 and the main exchange unit 20 are orthogonally connected by placement positions, that is, the plane where the blade unit 10 is located (a direction parallel to the paper in FIG. 1) and the plane where the main exchange unit 20 is located (as shown in FIG. 1) are perpendicular to each other.
  • This orthogonal connection method can flexibly support different types of connectors and adopt multiple protocols. It not only supports high bandwidth, but also facilitates the expansion of the blade unit 10 and the main Exchange unit 20.
  • the blade unit 10 and the main switching unit 20 are directly connected through the connector 30, and the backplane connection is no longer used, which simplifies the design of the server, so that the bandwidth of the connector is no longer limited by the backplane, and the High-speed, large-bandwidth connection.
  • the connector 30 between the blade unit 10 and the main switching unit 20 is replaced, the interface bandwidth or the communication interface protocol can be replaced without affecting the server frame, which greatly increases the flexibility of the internal interface of the system.
  • the front insertion mechanism includes a plurality of blade units 10, and the plurality of blade units 10 are arranged in an array.
  • the front insertion mechanism includes 16 blade units, and the 16 blade units are arranged in a 2 ⁇ 8 array, that is, 2 rows and 8 columns, in which 8 blade units in each row are arranged along the first direction perpendicular to the paper surface of FIG. 1 .
  • the front insertion mechanism includes 8 blade units 10, as shown in FIG. Layout in one direction.
  • the front insertion mechanism may include 6-16 blade units, or other numbers of blade units.
  • the blade unit 10 includes but is not limited to components such as a processor chip, a memory, a hard disk, or an expansion unit to provide computing and storage capabilities.
  • the number of main exchange units 20 is multiple, and the plurality of main exchange units 20 are stacked up and down, that is, arranged in a row along a second direction (up and down direction in FIG. 1) perpendicular to the first direction.
  • the main switching unit 20 is used to provide internal switching bandwidth and I/O interface expansion of the server.
  • the connector 30 may be an electrical connector or an optical connector. Moreover, the number of pins and signal rate of the connector can be flexibly expanded.
  • each blade unit 10 has a plate shape and extends along a vertical plane parallel to the paper surface of FIG. 1
  • each main exchange unit 20 has a plate shape and extends along a horizontal plane perpendicular to the paper surface of FIG. extend.
  • the connectors 30 are arranged in the same plane, which is parallel to the first direction and the second direction described above (ie, perpendicular to the paper surface of FIG. 1 ).
  • the number of connectors 30 connected to each blade unit 10 is at least two, and the number of connectors 30 connected to each main switching unit 20 is at least two.
  • the server includes a front insertion mechanism and a rear insertion mechanism.
  • the front insertion mechanism includes a blade unit 10
  • the rear insertion mechanism includes a main exchange unit 20.
  • the blade unit 10 and the main exchange unit 20 are connected to each other through a connector 30. Orthogonal connection.
  • the connector 30 adopts an orthogonal connector, that is, the blade unit 10 and the main switching unit 20 are orthogonally connected through the orthogonal connector.
  • the orthogonal connector can be an electrical connector or an optical connector.
  • the number of pins and signal rate of the quadrature connector can be flexibly expanded.
  • the blade unit 10 and the main switching unit 20 are directly connected through the orthogonal connector, and the backplane connection is no longer used, which simplifies the design of the server, and the bandwidth of the orthogonal connector is no longer limited by the backplane. Can obtain ultra-high speed, ultra-large bandwidth connection.
  • the interface bandwidth or the communication interface protocol can be changed without affecting the server frame, which greatly increases the flexibility of the internal interface of the system.
  • the front insertion mechanism includes a plurality of blade units 10, and the plurality of blade units 10 are arranged in an array.
  • the front insertion mechanism includes 16 blade units, and the 16 blade units are arranged in a 2 ⁇ 8 array.
  • the front insertion mechanism includes 8 blade units, and the 8 blade units are arranged in a 1 ⁇ 8 array, as shown in FIG. 2.
  • the front insertion mechanism may include any number of blade units 10, for example, the front insertion mechanism includes any number of blade units 10 from 6 to 16, or includes other numbers of blade units.
  • the blade unit 10 includes but is not limited to components such as a processor chip, a memory, a hard disk, or an expansion unit to provide computing and storage capabilities.
  • the main switching unit 20 includes a plurality of main switching nodes 21, and the plurality of main switching nodes 21 are stacked on top of each other.
  • the main switching unit 20 includes two main switching nodes 21, and the two main switching nodes 21 are stacked, as shown in FIG. 3.
  • the main switching unit 20 is used to provide internal switching bandwidth and I/O interface expansion of the server.
  • the server includes a front insertion mechanism and a rear insertion mechanism.
  • the front insertion mechanism includes a blade unit 10
  • the rear insertion mechanism includes a main exchange unit 20 and a cooling unit 40.
  • the blade unit 10 and the main switching unit 20 are connected in an orthogonal manner by a connector 30.
  • the blade unit 10 and the main switching unit 20 can be orthogonally connected through an orthogonal connector, or the blade unit 10 and the main switching unit 20 can be orthogonally connected in an orthogonal arrangement.
  • the connection of the blade unit 10 and the main switching unit 20 in an orthogonal manner not only realizes high-bandwidth transmission, but also facilitates the expansion of the blade unit 10 and the main switching unit 20.
  • the blade unit 10 and the main switching unit 20 are directly connected through the connector 30, and the backplane connection is no longer used, which simplifies the design of the server, so that the bandwidth of the connector is no longer limited by the backplane, and the High-speed, large-bandwidth connection.
  • the connector 30 between the blade unit 10 and the main switching unit 20 is replaced, the interface bandwidth or the communication interface protocol can be replaced without affecting the server frame, which greatly increases the flexibility of the internal interface of the system.
  • each front insertion mechanism may include 6-16 blade units, and multiple blade units 10 are arranged in an array.
  • the front insertion mechanism includes 16 blade units, and the 16 blade units are arranged in a 2 ⁇ 8 array.
  • the front insertion mechanism includes 8 blade units, and the 8 blade units are arranged in a 1 ⁇ 8 array, as shown in FIG. 2.
  • the blade unit 10 includes but is not limited to components such as a processor chip, a memory, a hard disk, or an expansion unit.
  • the main switching unit 20 includes a plurality of main switching nodes 21, and the plurality of main switching nodes 21 are stacked on top of each other.
  • the main switching unit 20 includes two main switching nodes 21, and the two main switching nodes 21 are stacked, as shown in FIG. 3.
  • the orthogonal connector may be an electrical connector or an optical connector. It is not difficult to understand that the number of pins and signal rate of the connector 30 can be flexibly expanded.
  • the cooling unit 40 is superimposed on the main switching unit 20 to cool the heat emitted by the blade unit 10 and the main switching unit 20, reduce the temperature of the server, and support high-power chips with higher heat consumption and reduce the server's PUE (power usage efficiency).
  • the cooling unit 40 includes an air-cooled module 41, wherein the air-cooled module 41 and the main exchange unit 20 are arranged one above the other, and the air-cooled module 41 may adopt one set of fans or multiple sets of fans.
  • the cooling unit 40 includes a liquid cooling module 42.
  • the liquid cooling module 42 includes a water inlet and outlet plug, a water divider 421, a water receiving tray 422, a leakage detector (not shown in the figure) and a leakage drain (not shown in the figure) ).
  • the drain pan 422 is arranged below the water trap 421, the leakage detector is arranged inside the drain pan 422, and the leakage drain is in communication with the drain pan 422.
  • the water separator 421 discharges the heat generated by the blade unit 10 and the main exchange unit 20, the water receiving tray 422 is used to collect the leakage of the water separator 421, and the leakage detector is used to detect whether there is any leakage of the water separator 421.
  • the drain pan 422 When the water separator 421 leaks, the drain pan 422 will collect the leakage. When the leakage detector detects the leakage, it will send out a leakage reminder message. At the same time, the leak exporter can be activated to remove the leakage in the drain pan 422. The liquid is drained to prevent leakage of liquid from damaging the boards inside the server.
  • the water inlet and outlet plugs are blind plugs. As shown in FIG. 4, the blind plug-in water inlet and outlet plugs are plugged into the water inlet pipe 423 and the water outlet pipe 424.
  • the water inlet pipe 423 and the water outlet pipe 424 are fixed to the rear of the machine frame of the server, and are connected to the internal water divider 421.
  • the blind plug is not only simple in design, but also easy to plug and maintain.
  • the server includes a front insertion mechanism and a rear insertion mechanism.
  • the front insertion mechanism includes a blade unit 10
  • the rear insertion mechanism includes a main exchange unit 20 and a cooling unit 40.
  • the blade unit 10 and the main switching unit 20 are connected in an orthogonal manner by a connector 30.
  • the blade unit 10 and the main switching unit 20 can be orthogonally connected through an orthogonal connector, or the blade unit 10 and the main switching unit 20 can be orthogonally connected in an orthogonal arrangement.
  • the connection of the blade unit 10 and the main switching unit 20 in an orthogonal manner not only realizes high-bandwidth transmission, but also facilitates the expansion of the blade unit 10 and the main switching unit 20.
  • each front insertion mechanism may include 6-16 blade units, and multiple blade units 10 are arranged in an array.
  • the front insertion mechanism includes 16 blade units, and the 16 blade units are arranged in a 2 ⁇ 8 array.
  • the front insertion mechanism includes 8 blade units, and the 8 blade units are arranged in a 1 ⁇ 8 array, as shown in FIG. 2.
  • the blade unit 10 includes but is not limited to components such as a processor chip, a memory, a hard disk, or an expansion unit.
  • the main switching unit 20 includes a plurality of main switching nodes 21, and the plurality of main switching nodes 21 are stacked on top of each other.
  • the main switching unit 20 includes two main switching nodes 21, and the two main switching nodes 21 are stacked, as shown in FIG. 3.
  • the orthogonal connector may be an electrical connector or an optical connector. It is not difficult to understand that the number of pins and signal rate of the connector 30 can be flexibly expanded.
  • the cooling unit 40 is superimposed on the main switching unit 20 to cool the heat emitted by the blade unit 10 and the main switching unit 20, reduce the temperature of the server, and support high-power chips with higher heat consumption and reduce the server's PUE (power usage efficiency).
  • the cooling unit 40 includes an air-cooled module 41 and a liquid-cooled module 42.
  • the air-cooled module 41 and the liquid-cooled module 42 are used to dissipate heat from the server, and the heat generated by the blade unit 10, the main switching unit 20 and the power supply is discharged .
  • the combination of air cooling and liquid cooling can improve heat dissipation efficiency, solve the bottleneck of the server's heat dissipation, and increase the server's processing capacity and switching bandwidth.
  • the liquid cooling module 42 includes a water inlet and outlet plug, a water distributor 421, a water receiving tray 422, a leakage detector (not shown in the figure), and a leakage exporter (not shown in the figure). Below the water tank 421, a liquid leakage detector is arranged inside the water receiving pan 422, and the leakage liquid exporter is in communication with the water receiving pan 422.
  • the water separator 421 discharges the heat generated by the blade unit 10 and the main exchange unit 20, the water receiving tray 422 is used to collect the leakage of the water separator 421, and the leakage detector is used to detect whether there is any leakage of the water separator 421.
  • the drip tray 422 will collect the leaking liquid.
  • the leak detector detects the leak, it will send out a leak reminder message.
  • the leak exporter can be activated to remove the leak in the drip tray 422. The liquid is drained to prevent leakage of liquid from damaging the boards inside the server.
  • the water inlet and outlet plugs are blind plugs.
  • the blind plug-in water inlet and outlet plugs are connected to the water inlet pipe 423 and the water outlet pipe 424.
  • the water inlet pipe 423 and the water outlet pipe 424 are fixed to the rear of the machine frame of the server, and are connected to the internal water divider 421.
  • the blind plug is not only simple in design, but also easy to plug and maintain.
  • the server includes a front insertion mechanism and a rear insertion mechanism.
  • the front insertion mechanism includes a blade unit 10
  • the rear insertion mechanism includes a main exchange unit 20 and a cooling unit 40.
  • the blade unit 10 and the main switching unit 20 are connected in an orthogonal manner by a connector 30.
  • the blade unit 10 and the main switching unit 20 can be orthogonally connected through an orthogonal connector, or the blade unit 10 and the main switching unit 20 can be orthogonally connected in an orthogonal arrangement.
  • the connection of the blade unit 10 and the main switching unit 20 in an orthogonal manner not only realizes high-bandwidth transmission, but also facilitates the expansion of the blade unit 10 and the main switching unit 20.
  • the blade unit 10 and the main switching unit 20 are directly connected through the connector 30, and the backplane connection is no longer used, which simplifies the design of the server, so that the bandwidth of the connector is no longer limited by the backplane, and the High-speed, large-bandwidth connection.
  • the connector 30 between the blade unit 10 and the main switching unit 20 is replaced, the interface bandwidth or the communication interface protocol can be replaced without affecting the server frame, which greatly increases the flexibility of the internal interface of the system.
  • each front insertion mechanism may include 6-16 blade units, and multiple blade units 10 are arranged in an array.
  • the front insertion mechanism includes 16 blade units, and the 16 blade units are arranged in a 2 ⁇ 8 array.
  • the front insertion mechanism includes 8 blade units, and the 8 blade units are arranged in a 1 ⁇ 8 array, as shown in FIG. 2.
  • the blade unit 10 includes but is not limited to components such as a processor chip, a memory, a hard disk, or an expansion unit.
  • the main switching unit 20 includes a plurality of main switching nodes 21, and the plurality of main switching nodes 21 are stacked on top of each other.
  • the main switching unit 20 includes two main switching nodes 21, and the two main switching nodes 21 are stacked, as shown in FIG. 3.
  • the orthogonal connector may be an electrical connector or an optical connector. It is not difficult to understand that the number of pins and signal rate of the connector 30 can be flexibly expanded.
  • the cooling unit 40 is superimposed on the main switching unit 20 to cool the heat emitted by the blade unit 10 and the main switching unit 20, reduce the temperature of the server, and support high-power chips with higher heat consumption and reduce the server's PUE (power usage efficiency).
  • the cooling unit 40 includes an air-cooled module 41 and a liquid-cooled module 42.
  • the air-cooled module 41 and the liquid-cooled module 42 are used to dissipate heat from the server, and the heat generated by the blade unit 10, the main switching unit 20 and the power supply is discharged .
  • the combination of air cooling and liquid cooling can improve heat dissipation efficiency, solve the bottleneck of the server's heat dissipation, and increase the server's processing capacity and switching bandwidth.
  • the liquid cooling module 42 includes a water inlet and outlet plug, a water distributor 421, a water receiving tray 422, a leakage detector (not shown in the figure), and a leakage exporter (not shown in the figure). Below the water tank 421, a liquid leakage detector is arranged inside the water receiving pan 422, and the leakage liquid exporter is in communication with the water receiving pan 422.
  • the water separator 421 discharges the heat generated by the blade unit 10 and the main exchange unit 20, the water receiving tray 422 is used to collect the leakage of the water separator 421, and the leakage detector is used to detect whether there is any leakage of the water separator 421.
  • the drip tray 422 will collect the leaking liquid.
  • the leak detector detects the leak, it will send out a leak reminder message.
  • the leak exporter can be activated to remove the leak in the drip tray 422. The liquid is drained to prevent leakage of liquid from damaging the boards inside the server.
  • the water inlet and outlet plugs are blind plugs. As shown in FIG. 4, the blind plug-in water inlet and outlet plugs are plugged into the water inlet pipe 423 and the water outlet pipe 424.
  • the water inlet pipe 423 and the water outlet pipe 424 are fixed to the rear of the machine frame of the server, and are connected to the internal water divider 421.
  • the blind plug is not only simple in design, but also easy to plug and maintain.
  • the liquid cooling module 42 can be set in the processor and chip position with high power consumption in the server, the liquid cooling module is used to dissipate heat from the high power consumption chip and the processor, and the air cooling module 41 is set in For chips and processor locations with relatively low power consumption, the air-cooled module 41 is used to dissipate heat from the chips and processor locations with relatively low power consumption. It is not difficult to understand that the cooling unit 40 of this embodiment may choose to use an air-cooled module 41 or a liquid-cooled module 42 according to actual use conditions, or choose to use a combination of the air-cooled module 41 and the liquid-cooled module 42.
  • the rear insertion mechanism further includes a power supply 50 and a chassis management unit 60.
  • the power supply 50 and the chassis management unit 60 are respectively connected to the front insertion mechanism through connectors fixed on the backplane 80 .
  • the power supply 50 is used to provide electrical energy to meet the normal operation of the server.
  • the frame management unit 60 is used to manage the units and modules of the server, so that the server can run efficiently and reasonably.
  • the power supply 50 and the chassis management unit 60 are electrically connected to the front insertion mechanism through connectors fixed on the backplane 80, respectively.
  • a connector is fixed on the backplane 80, and the power supply 50 is electrically connected to the front insertion mechanism through the connector.
  • one end of the two insertion ends of the connector is connected to the power supply 50, and the other end is connected to the front insertion mechanism to realize the power supply.
  • 50 Electrical connection with the front insertion mechanism.
  • the frame management unit 60 is electrically connected to the front insertion mechanism through a connector.
  • one end of the two insertion ends of the connector is inserted into the frame management unit 60, and the other end is inserted into the front insertion mechanism, so that the frame management unit 60 is connected to the front insertion mechanism.
  • the power supply 50 and the chassis management unit 60 can also be connected to other modules or units through connectors fixed on the backplane 80.
  • the connector can select connectors with different functions according to the connection object to realize the corresponding signal connection.
  • the power supply 50 includes a main power supply and an auxiliary power supply.
  • the main power supply provides power to the server;
  • the auxiliary power supply provides a backup power supply for the server, that is, the redundancy of the power supply 50 is increased to improve the reliability of the power supply 50.
  • the power supply 50 can be supplied in a redundant manner of n+1 or n+n.
  • the power supply 50 and the chassis management unit 60 communicate in a point-to-point or bus manner, and the chassis management unit 60 queries the state of the power supply 50 to perform operations such as power saving and firmware upgrade.
  • the server includes a front insertion mechanism and a rear insertion mechanism.
  • the front insertion mechanism includes a blade unit 10
  • the rear insertion mechanism includes a main exchange unit 20, an auxiliary exchange unit 70, and a cooling unit 40.
  • 10 and the main switching unit 20 are connected in an orthogonal manner by a connector 30.
  • the blade unit 10, the auxiliary exchange unit 70, and the cooling unit 40 are fixed on the back plate 80, and the blade unit 10 and the auxiliary exchange unit 70 are signally connected by connectors.
  • a through hole (not shown in the figure) is provided on the back plate 80, the connector is embedded in the through hole of the back plate 80, and the blade unit 10 and the auxiliary exchange unit 70 are inserted into the connector.
  • the backplane 80 only provides the connection of power distribution and control management signals, and the signal connection between the auxiliary switching unit 70 and the blade unit 10, but the backplane 80 does not provide the signal connection between the blade unit 10 and the main switching unit 20.
  • the blade unit 10 and the main switching unit 20 are connected in an orthogonal manner through the connector 30, which can flexibly support different types of connectors, not only supports high bandwidth, but also facilitates the expansion of the blade unit 10 and the main switching unit 20.
  • the blade unit 10 and the auxiliary switching unit 70 are connected through the backplane 80, which increases the flexibility of server configuration.
  • the blade unit 10 and the main switching unit 20 are directly connected through the connector 30, and the backplane connection is no longer used, which simplifies the design of the server, so that the bandwidth of the connector is no longer limited by the backplane, and the High-speed, large-bandwidth connection.
  • the connector 30 between the blade unit 10 and the main switching unit 20 is replaced, the interface bandwidth or the communication interface protocol can be replaced without affecting the server frame, which greatly increases the flexibility of the internal interface of the server.
  • each front insertion mechanism may include 6-16 blade units, and multiple blade units 10 are arranged in an array.
  • the front insertion mechanism includes 16 blade units, and the 16 blade units are arranged in a 2 ⁇ 8 array.
  • the front insertion mechanism includes 8 blade units, and the 8 blade units are arranged in a 1 ⁇ 8 array, as shown in FIG. 2.
  • the blade unit 10 includes but is not limited to components such as a processor chip, a memory, a hard disk, or an expansion unit.
  • the main switching unit 20 includes a plurality of main switching nodes 21, and the plurality of main switching nodes 21 are stacked on top of each other.
  • the main switching unit 20 includes two main switching nodes 21, and the two main switching nodes 21 are stacked, as shown in FIG. 5.
  • the connector 30 is an orthogonal connector, which may be an electrical connector or an optical connector. It is not difficult to understand that the number of pins and signal rate of the connector 30 can be flexibly expanded.
  • the cooling unit 40 is superimposed on the main switching unit 20 to cool the heat emitted by the blade unit 10 and the main switching unit 20, reduce the temperature of the server, and support high-power chips with higher heat consumption and reduce the server's PUE (power usage efficiency).
  • the cooling unit 40 includes an air-cooled module 41, wherein the air-cooled module 41 and the main exchange unit 20 are arranged one above the other, and the air-cooled module 41 may adopt one set of fans or multiple sets of fans.
  • the cooling unit 40 includes a liquid cooling module 42.
  • the liquid cooling module 42 includes a water inlet and outlet plug, a water distributor 421, a water receiving tray 422, a leakage detector (not shown in the figure), and a leakage exporter (not shown in the figure).
  • a liquid leakage detector is arranged inside the water receiving pan 422, and the leakage liquid exporter is in communication with the water receiving pan 422.
  • the water separator 421 discharges the heat generated by the blade unit 10 and the main exchange unit 20, the water receiving tray 422 is used to collect the leakage of the water separator 421, and the leakage detector is used to detect whether there is any leakage of the water separator 421.
  • the drip tray 422 When the water separator 421 leaks, the drip tray 422 will collect the leaking liquid. When the leak detector detects the leak, it will send out a leak reminder message. At the same time, the leak exporter can be activated to remove the leak in the drip tray 422. The liquid is drained to prevent leakage of liquid from damaging the boards inside the server.
  • the water inlet and outlet plugs are blind plugs. As shown in FIG. 4, the blind plug-in water inlet and outlet plugs are plugged into the water inlet pipe 423 and the water outlet pipe 424.
  • the water inlet pipe 423 and the water outlet pipe 424 are fixed to the rear of the machine frame of the server, and are connected to the internal water divider 421.
  • the blind plug is not only simple in design, but also easy to plug and maintain.
  • the cooling unit 40 includes an air-cooled module 41 and a liquid-cooled module 42.
  • the air-cooled module 41 and the liquid-cooled module 42 are used to dissipate heat from the server, and the heat generated by the blade unit 10, the main switching unit 20 and the power supply is discharged .
  • the combination of air cooling and liquid cooling can improve heat dissipation efficiency, solve the bottleneck of the server's heat dissipation, and increase the server's processing capacity and switching bandwidth.
  • the liquid cooling module 42 can be set in the processor and chip position with high power consumption in the server, the liquid cooling module is used to dissipate heat from the high power consumption chip and the processor, and the air cooling module 41 is set in For chips and processor locations with relatively low power consumption, the air-cooled module 41 is used to dissipate heat from the chips and processor locations with relatively low power consumption.
  • an auxiliary exchange unit 70 may be provided at the position of the liquid-cooled module 42 to increase The bandwidth of the server.
  • the rear insertion mechanism further includes a power supply 50 and a chassis management unit 60, and the power supply 50 and the chassis management unit 60 are respectively electrically connected to the front insertion mechanism.
  • the power supply 50 is used to provide electrical energy to meet the normal operation of the server.
  • the frame management unit 60 is used to collect and manage the basic information of the components in the frame, power off the board, various sensor information, fan speed reading and adjustment, liquid leakage detection, power supply monitoring, firmware loading and update operations, etc., to enable the server Efficient and reasonable operation.
  • the power supply 50 and the chassis management unit 60 are electrically connected to the front insertion mechanism through connectors fixed on the backplane 80, respectively.
  • a connector is fixed on the backplane 80, and the power supply 50 is electrically connected to the front insertion mechanism through the connector.
  • one end of the two insertion ends of the connector is connected to the power supply 50, and the other end is inserted into the front insertion mechanism to realize the power supply.
  • 50 Electrical connection with the front insertion mechanism.
  • the frame management unit 60 is electrically connected to the front insertion mechanism through a connector.
  • one end of the two insertion ends of the connector is inserted into the frame management unit 60, and the other end is inserted into the front insertion mechanism, so that the frame management unit 60 is connected to the front insertion mechanism.
  • the power supply 50 and the chassis management unit 60 can also be connected to other modules or units through a connector fixed on the backplane 80.
  • the connector can select connectors with different functions according to the connection object to realize the corresponding signal connection.
  • the power supply 50 includes a main power supply and an auxiliary power supply.
  • the main power supply provides electric energy for the server; the auxiliary power supply provides a backup power supply for the server to improve the reliability of the power supply 50.
  • the chassis management unit 60 includes a main chassis management unit and an auxiliary chassis management unit, and the auxiliary chassis management unit is a backup management unit of the main chassis management unit. It is not difficult to understand that the main frame management unit and the auxiliary frame management unit can be the masters of each other, as long as the frame management unit 60 can serve normally.
  • the server provided in this embodiment can be used for a hyper-converged server, and can also be used for other servers that require high integration and high processing capabilities.
  • the blade unit and the main switching unit are connected in an orthogonal manner through a connector, which can increase the interface bandwidth between the blade unit and the main switching unit, improve the transmission efficiency, and make the backplane separate from the high-speed signal to back
  • a connector which can increase the interface bandwidth between the blade unit and the main switching unit, improve the transmission efficiency, and make the backplane separate from the high-speed signal to back

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Power Engineering (AREA)
  • Human Computer Interaction (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Computer Hardware Design (AREA)
  • Cooling Or The Like Of Electrical Apparatus (AREA)

Abstract

本公开提供一种服务器,其包括:前插机构和后插机构,所述前插机构包括刀片单元,所述后插机构包括主交换单元,所述刀片单元和所述主交换单元通过连接器以正交方式连接。

Description

服务器
相关申请的交叉引用
本申请要求2020年6月11日提交的中国专利申请CN202010531347.4的优先权,该申请的全部内容以引用的方式并入本文。
技术领域
本公开涉及计算机技术领域,尤其涉及一种服务器。
背景技术
刀片服务器以高集成度、高处理能力、共用散热和电源等特点,在电信、超算等领域得到广泛应用。在传统刀片服务器架构中,刀片单元和交换单元是通过背板实现信号连接的,因此,信号速率和信号带宽很大程度上依赖于背板。
背景技术
刀片服务器以高集成度、高处理能力、共用散热和电源等特点,在电信、超算等领域得到广泛应用。在传统刀片服务器架构中,刀片单元和交换单元是通过背板实现信号连接的,因此,信号速率和信号带宽很大程度上依赖于背板。
然而,随着通讯协议的更迭以及信号传输速率的提升,部分通讯协议需要支持56G PAM4和112G PAM4,对信号衰减和串扰的要求越来越高,传统的背板难以支撑如此高的信号传输速率。而且,背板设计完成后,信号带宽就基本确定,不容易对带宽进行升级。
发明内容
本公开实施例提出一种服务器,旨在提高服务器的带宽和传输速率,以及提高内部接口的灵活性。
本公开实施例提供的服务器包括:前插机构和后插机构,所述前插机构包括刀片单元,所述后插机构包括主交换单元,所述刀片单元和所述主交换单元通过连接器以正交方式连接。
附图说明
图1为本公开实施例提供的服务器的侧视图;
图2为本公开实施例提供的前插机构的主视图;
图3为本公开实施例提供的后插机构的后视图;
图4为本公开实施提供的液冷模块的部分结构示意图;
图5为本公开另一实施例提供的后插机构的后视图;
图6为本公开另一实施例提供的服务器的侧视图。
附图标记说明:
10-刀片单元,20-主交换单元,30-连接器,40-冷却单元,41-风冷模块,42-液冷模块,421-分水器,422-接水盘,423-进水管,424-出水管,50-电源,60-机框管理单元,70-辅助交换单元,80-背板。
具体实施方式
为使本领域的技术人员更好地理解本公开的技术方案,下面结合附图对本公开提供的服务器进行详细描述。
在下文中将参考附图更充分地描述示例实施例,但是所述示例实施例可以以不同形式来体现且不应当被解释为限于本文阐述的实施例。反之,提供这些实施例的目的在于使本公开透彻和完整,并将使本领域技术人员充分理解本公开的范围。
如本文所使用的,术语“和/或”包括一个或多个相关列举条目的任何和所有组合。
本文所使用的术语仅用于描述特定实施例,且不意欲限制本公开。如本文所使用的,单数形式“一个”和“该”也意欲包括复数形式,除非上下文另外清楚指出。还将理解的是,当本说明书中使用术语“包括”和/或“由……制成”时,指定存在所述特征、整体、步骤、操作、元件和/或组件,但不排除存在或添加一个或多个其它特征、整体、步骤、操作、元件、组件和/或其群组。
本文所述实施例可借助本公开的理想示意图而参考平面图和/或截面图进行描述。因此,可根据制造技术和/或容限来修改示例图示。因此,实施例不限于附图中所示的实施例,而是包括基于制造工艺而形成的配置的修改。因此,附图中例示的区具有示意性属性,并且图中所示区的形状例示了元件的区的具体形状,但并不旨在是限制性的。
除非另外限定,否则本文所用的所有术语(包括技术和科学术语)的含义与本领域普通技术人员通常理解的含义相同。还将理解,诸如那些在常用字典中限定的那些术语应当被解释为具有与其在相关技术以及本公开的背景下的含义一致的含义,且将不解释为具有理想化或过度形式上的含义,除非本文明确如此限定。
服务器包括多个刀片单元和交换单元,在相关技术中,刀片单元和交换单元通过背板实现高速信号传递。为了有效利用服务器空间和降低成本,需要将刀片单元集成在一起。然而,由于背板的限制,刀片单元和交换单元之间的接口带宽有限,而且背板完成设计后,无法进行扩容。
本公开实施例提供一种服务器,既可将刀片单元集成在一起,又可提高刀片单元和交换单元之间的接口带宽,而且可以对服务器进行扩容。
实施例一
图1为本公开实施例提供的一种服务器的侧视图,图2为图1中的服务器的前插机构的主视图。参照图1,服务器包括前插机构和后插机构,其中,前插机构包括刀片单元10,后插机构包括主交换单元20,而且,刀片单元10和主交换单元20通过连接器30以正交方式连接。
在本实施例中,刀片单元10和主交换单元20通过摆放位置实现正交连接,即刀片单元10所在平面(与图1的纸面平行的方向)与主交换单元20所在平面(与图1的水平方向平行的方向)相互垂直,这种通过摆放位置实现正交连接方式可以灵活地支持不同类型连接器,以及采用多种协议,不仅支持高带宽,而且方便扩展刀片单元10和主交换单元20。
在本实施例中,刀片单元10和主交换单元20直接通过连接器30连接,不再采用背板连接,简化了服务器的设计,使得连接器的带宽不再受背板的限制,可获得超高速率、超大带宽的连接。另外,当更换刀片单元10和主交换单元20之间的连接器30时,可在不影响服务器的框架下更换接口带宽或通讯接口协议,大大增加系统内部接口的灵活性。
在一些实施方式中,前插机构包括多个刀片单元10,多个刀片单元10采用阵列方式设置。例如,前插机构包括16个刀片单元,16个刀片单元以2×8阵列设置,即,2行8列,其中每行的8个刀片单元沿与图1的纸面垂直的第一方向布置。再如,前插机构包括8个刀片单元10,如图2所示,8个刀片单元10以1×8阵列设置,即,一行的8个刀片单元10沿与图1的纸面垂直的第一方向布置。实际上,前插机构可以包括6-16个刀片单元,或者包括其他数量的刀片单元。刀片单元10包括但不限于处理器芯片、内存、硬盘或扩展单元等部件,用以提供计算及存储能力。
在一些实施方式中,主交换单元20的数量为多个,多个主交换单元20上下叠置,即沿与第一方向垂直的第二方向(图1中的上下方向)布置成列。例如,主交换单元20为2个,两个主交换单元20沿上下方向叠置,如图3所示。主交换单元20用于提供服务器内部交换带宽和I/O接口扩展。
在本实施例中,连接器30可以是电连接器,也可以是光连接器。而且,连接器的管脚数和信号速率可灵活扩展。
在一些实施方式中,每个刀片单元10呈板状,沿与图1的纸面平行的竖直平面延伸,且每个主交换单元20呈板状,沿与图1的纸面垂直的水平面延伸。
在一些实施方式中,如图1所示,各个连接器30布置在同一平面中,该平面与上述第一方向和第二方向均平行(即,与图1的纸面垂直)。
在一些实施方式中,与每个刀片单元10连接的连接器30的数量为至少两个,与每个主交换单元20连接的连接器30的数量为至少两个。
实施例二
如图1所示,服务器包括前插机构和后插机构,其中,前插机构包括刀片单元10,后插机构包括主交换单元20,而且,刀片单元10和主交换单元20通过连接器30以正交方式连接。
在本实施例中,连接器30采用正交连接器(orthogonal connector),即刀片单元10和主交换单元20是通过正交连接器实现正交连接。其中,正交连接器可以是电连接器,也可以是光连接器。另外,正交连接器的管脚数和信号速率可灵活扩展。
在本实施例中,刀片单元10和主交换单元20直接通过正交连接器连接,不再采用背板连接,简化了服务器的设计,而且正交连接器的带宽不再受背板的限制,可获得超高速率、超大带宽的连接。另外,在更换刀片单元10和主交换单元20之间的正交连接器时,可在不影响服务器的框架下更换接口带宽或通讯接口协议,大大增加系统内部接口的灵活性。
在本实施例中,前插机构包括多个刀片单元10,而且多个刀片单元10采用阵列方式设置。例如,前插机构包括16个刀片单元,16个刀片单元以2×8阵列设置。再如,前插机构包括8个刀片单元,8个刀片单元以1×8阵列设置,如图2所示。实际上,前插机构可以包括任意数量的刀片单元10,例如前插机构包括6-16中任意数量的刀片单元10,或者包括其他数量的刀片单元。刀片单元10包括但不限于处理器芯片、内存、硬盘或扩展单元等部件,用以提供计算及存储能力。
在一些实施方式中,主交换单元20包括多个主交换节点21,多个主交换节点21上下叠置。例如,主交换单元20包括两个主交换节点21,两个主交换节点21叠置,如图3所示。主交换单元20用于提供服务器内部交换带宽和I/O接口扩展。
实施例三
如图1和图3所示,在本实施例中,服务器包括前插机构和后插机构,其中,前插机构包括刀片单元10,后插机构包括主交换单元20和冷却单元40,刀片单元10和主交换单元20通过连接器30以正交方式连接。在本实施例中,刀片单元10和主交换单元20可以通过正交连接器实现正交连接,或者,刀片单元10和主交换单元20以正交摆放方式实现正交连接。刀片单元10和主交换单元20以正交方式连接不仅实现了高带宽传输,而且方便扩展刀片单元10和主交换单元20。
在本实施例中,刀片单元10和主交换单元20直接通过连接器30连接,不再采用背板连接,简化了服务器的设计,使得连接器的带宽不再受背板的限制,可获得超高速率、超大带宽的连接。另外,当更换刀片单元10和主交换单元20之间的连接器30时,可在不影响服务器的框架下更换接口带宽或通讯接口协议,大大增加系统内部接口的灵活性。
在一些实施方式中,每个前插机构可以包括6-16个刀片单元,多个刀片单元10采用阵列方式设置。例如,前插机构包括16个刀片单元,16个刀片单元以2×8阵列设置。再如,前插机构包括8个刀片单元,8个刀片单元以1×8阵列设置,如图2所示。刀片单元10包括但不限于处理器芯片、内存、硬盘或扩展单元等部件。在一些实施方式中,主交换单元20包括多个主交换节点21,多个主交换节点21上下叠置。例如,主交换单元20包括两个主交换节点21,两个主交换节点21叠置,如图3所示。在一些实施方式中,正交连接器可以是电连接器,也可以是光连接器。不难理解,连接器30的管脚数和信号速率可灵活扩展。
在本实施例中,冷却单元40与主交换单元20叠置,用于冷却刀片单元10和主交换单元20散发的热量,降低服务器的温度,以支持更高热耗的大功率芯片,降低服务器的PUE(电源使用效率)。
在一些实施方式中,冷却单元40包括风冷模块41,其中,风冷模块41与主交换单元20上下叠置设置,风冷模块41可以采用一组风扇或多组风扇。
在一些实施方式中,冷却单元40包括液冷模块42。如图3和图4所示,液冷模块42包括进出水插头、分水器421、接水盘422、漏液检测器(图中未示出)和漏液导出器(图中未示出)。接水盘422设置在分水器421的下方,漏液检测器设置在接水盘422的内侧,漏液导出器与接水盘422连通。分水器421将刀片单元10和主交换单元20产生的热量排出,接水盘422用于收集分水器421的漏液,漏液检测器用于检测分水器421是否存在漏液。当分水器421发生漏液时,接水盘422会聚集漏液,漏液检测器检测到漏液时将发出漏液提醒信息, 同时可以启动漏液导出器,将接水盘422内的漏液排出,避免漏液损坏服务器内部的单板。
在一些实施方式中,进出水插头采用盲插式插头。如图4所示,盲插式进出水插头与进水管423和出水管424插接。进水管423和出水管424固定于服务器的机框的后部,并与内部的分水器421连接。盲插式插头不仅设计简洁,而且插拔维护方便。
实施例四
如图1和图3所示,在一些实施方式中,服务器包括前插机构和后插机构,其中,前插机构包括刀片单元10,后插机构包括主交换单元20和冷却单元40,刀片单元10和主交换单元20通过连接器30以正交方式连接。在本实施例中,刀片单元10和主交换单元20可以通过正交连接器实现正交连接,或者,刀片单元10和主交换单元20以正交摆放方式实现正交连接。刀片单元10和主交换单元20以正交方式连接不仅实现了高带宽传输,而且方便扩展刀片单元10和主交换单元20。
在本实施例中,刀片单元10和主交换单元20直接通过连接器30连接,不再采用背板连接,简化了服务器的设计,使得连接器的带宽不再受背板的限制,可获得超高速率、超大带宽的连接。另外,当更换刀片单元10和主交换单元20之间的连接器30时,可在不影响服务器的框架下更换接口带宽或通讯接口协议,大大增加系统内部接口的灵活性。
在一些实施方式中,每个前插机构可以包括6-16个刀片单元,多个刀片单元10采用阵列方式设置。例如,前插机构包括16个刀片单元,16个刀片单元以2×8阵列设置。再如,前插机构包括8个刀片单元,8个刀片单元以1×8阵列设置,如图2所示。刀片单元10包括但不限于处理器芯片、内存、硬盘或扩展单元等部件。在一些实施方式中,主交换单元20包括多个主交换节点21,多个主交换节点21上下叠置。例如,主交换单元20包括两个主交换节点21,两个主交换节点21叠置,如图3所示。在一些实施方式中,正交连接器可以是电连接器,也可以是光连接器。不难理解,连接器30的管脚数和信号速率可灵活扩展。
在本实施例中,冷却单元40与主交换单元20叠置,用于冷却刀片单元10和主交换单元20散发的热量,降低服务器的温度,以支持更高热耗的大功率芯片,降低服务器的PUE(电源使用效率)。
在一些实施方式中,冷却单元40包括风冷模块41和液冷模块42,利用风冷模块41和液冷模块42对服务器进行散热,将刀片单元10、主交换单元20和电源产生的热量排出。风冷和液冷结合的方式可以提高散热效率,解决服务器散热的瓶颈,提高服务器的处理能力和交换带宽。
其中,风冷模块41与主交换单元20上下叠置设置,风冷模块41可以采用一组风扇或多组风扇。液冷模块42包括进出水插头、分水器421、接水盘422、漏液检测器(图中未示出)和漏液导出器(图中未示出),接水盘422设置在分水器421的下方,漏液检测器设置在接水盘422的内侧,漏液导出器与接水盘422连通。分水器421将刀片单元10和主交换单元20产生的热量排出,接水盘422用于收集分水器421的漏液,漏液检测器用于检测分水器421是否存在漏液。当分水器421发生漏液时,接水盘422会聚集漏液,漏液检测器检测到漏液时将发出漏液提醒信息,同时可以启动漏液导出器,将接水盘422内的漏液排出,避免漏液损坏服务器内部的单板。
在一些实施方式中,进出水插头采用盲插式插头。如图4所示,盲插式进出水插头与进 水管423和出水管424插接。进水管423和出水管424固定于服务器的机框的后部,并与内部的分水器421连接。盲插式插头不仅设计简洁,而且插拔维护方便。
实施例五
如图1和图3所示,在一些实施方式中,服务器包括前插机构和后插机构,其中,前插机构包括刀片单元10,后插机构包括主交换单元20和冷却单元40,刀片单元10和主交换单元20通过连接器30以正交方式连接。在本实施例中,刀片单元10和主交换单元20可以通过正交连接器实现正交连接,或者,刀片单元10和主交换单元20以正交摆放方式实现正交连接。刀片单元10和主交换单元20以正交方式连接不仅实现了高带宽传输,而且方便扩展刀片单元10和主交换单元20。
在本实施例中,刀片单元10和主交换单元20直接通过连接器30连接,不再采用背板连接,简化了服务器的设计,使得连接器的带宽不再受背板的限制,可获得超高速率、超大带宽的连接。另外,当更换刀片单元10和主交换单元20之间的连接器30时,可在不影响服务器的框架下更换接口带宽或通讯接口协议,大大增加系统内部接口的灵活性。
在一些实施方式中,每个前插机构可以包括6-16个刀片单元,多个刀片单元10采用阵列方式设置。例如,前插机构包括16个刀片单元,16个刀片单元以2×8阵列设置。再如,前插机构包括8个刀片单元,8个刀片单元以1×8阵列设置,如图2所示。刀片单元10包括但不限于处理器芯片、内存、硬盘或扩展单元等部件。在一些实施方式中,主交换单元20包括多个主交换节点21,多个主交换节点21上下叠置。例如,主交换单元20包括两个主交换节点21,两个主交换节点21叠置,如图3所示。在一些实施方式中,正交连接器可以是电连接器,也可以是光连接器。不难理解,连接器30的管脚数和信号速率可灵活扩展。
在本实施例中,冷却单元40与主交换单元20叠置,用于冷却刀片单元10和主交换单元20散发的热量,降低服务器的温度,以支持更高热耗的大功率芯片,降低服务器的PUE(电源使用效率)。
在一些实施方式中,冷却单元40包括风冷模块41和液冷模块42,利用风冷模块41和液冷模块42对服务器进行散热,将刀片单元10、主交换单元20和电源产生的热量排出。风冷和液冷结合的方式可以提高散热效率,解决服务器散热的瓶颈,提高服务器的处理能力和交换带宽。
其中,风冷模块41与主交换单元20上下叠置设置,风冷模块41可以采用一组风扇或多组风扇。液冷模块42包括进出水插头、分水器421、接水盘422、漏液检测器(图中未示出)和漏液导出器(图中未示出),接水盘422设置在分水器421的下方,漏液检测器设置在接水盘422的内侧,漏液导出器与接水盘422连通。分水器421将刀片单元10和主交换单元20产生的热量排出,接水盘422用于收集分水器421的漏液,漏液检测器用于检测分水器421是否存在漏液。当分水器421发生漏液时,接水盘422会聚集漏液,漏液检测器检测到漏液时将发出漏液提醒信息,同时可以启动漏液导出器,将接水盘422内的漏液排出,避免漏液损坏服务器内部的单板。
在一些实施方式中,进出水插头采用盲插式插头。如图4所示,盲插式进出水插头与进水管423和出水管424插接。进水管423和出水管424固定于服务器的机框的后部,并与内部的分水器421连接。盲插式插头不仅设计简洁,而且插拔维护方便。
在一些实施方式中,可以将液冷模块42设置在服务器中功耗较高的处理器、芯片位置,利用液冷模块对高功耗的芯片和处理器进行散热,将风冷模块41设置在功耗相对较低的芯片、处理器位置,利用风冷模块41对功耗相对较低的芯片和处理器位置进行散热。不难理解,本实施例的冷却单元40可以根据实际使用情况,选择采用风冷模块41或液冷模块42,或者选择采用风冷模块41和液冷模块42相结合的方式。
在一些实施方式中,如图3所示,后插机构还包括电源50和机框管理单元60,电源50和机框管理单元60通过固定在背板80上的连接器分别与前插机构连接。其中,电源50用于提供电能,以满足服务器的正常运行。机框管理单元60用于管理服务器各单元和模块,使服务器高效合理的运行。
在一些实施方式中,电源50和机框管理单元60通过固定在背板80上的连接器分别与前插机构电连接。换言之,在背板80上固定有连接器,电源50与前插机构通过连接器电连接,例如,连接器的两个插接端一端插接电源50,另一端插接前插机构,实现电源50与前插机构的电连接。机框管理单元60与前插机构通过连接器电连接,例如,连接器的两个插接端一端插接,机框管理单元60,另一端插接前插机构,实现机框管理单元60与前插机构的电连接。在一些实施方式中,电源50和机框管理单元60还可以通过固定在背板80上的连接器与其他模块或单元连接。其中,连接器可以根据连接对象选择不同功能的连接器,以实现相应的信号连接。
在一些实施方式中,电源50包括主电源和辅助电源,主电源为服务器提供电能;辅助电源为服务器提供备用电源,即增加电源50的冗余,以提高电源50的可靠性。例如,电源50可以采用n+1或n+n的冗余方式供电。电源50与机框管理单元60之间采用点对点或总线方式进行通讯,由机框管理单元60对电源50的状态进行查询,以进行节电及固件升级等操作。
实施例六
如图5和图6所示,服务器包括前插机构和后插机构,其中,前插机构包括刀片单元10,后插机构包括主交换单元20、辅助交换单元70和冷却单元40,在刀片单元10和主交换单元20通过连接器30以正交方式连接。刀片单元10、辅助交换单元70和冷却单元40固定在背板80上,刀片单元10和辅助交换单元70通过连接器信号连接。例如,在背板80上设置通孔(图中未示出),将连接器嵌置于背板80的通孔内,刀片单元10和辅助交换单元70插接于连接器内。背板80仅提供电源分配和控制管理信号的连接,以及辅助交换单元70和刀片单元10的信号连接,但背板80不提供刀片单元10和主交换单元20的信号连接。
在本实施例中,刀片单元10和主交换单元20通过连接器30以正交方式连接,可以灵活地支持不同类型连接器,不仅支持高带宽,而且方便扩展刀片单元10和主交换单元20。同时,刀片单元10与辅助交换单元70通过背板80连接,增加了服务器配置的灵活性。
在本实施例中,刀片单元10和主交换单元20直接通过连接器30连接,不再采用背板连接,简化了服务器的设计,使得连接器的带宽不再受背板的限制,可获得超高速率、超大带宽的连接。另外,当更换刀片单元10和主交换单元20之间的连接器30时,可在不影响服务器的框架下更换接口带宽或通讯接口协议,大大增加服务器内部接口的灵活性。
在一些实施方式中,每个前插机构可以包括6-16个刀片单元,多个刀片单元10采用阵列方式设置。例如,前插机构包括16个刀片单元,16个刀片单元以2×8阵列设置。再如, 前插机构包括8个刀片单元,8个刀片单元以1×8阵列设置,如图2所示。刀片单元10包括但不限于处理器芯片、内存、硬盘或扩展单元等部件。在一些实施方式中,主交换单元20包括多个主交换节点21,多个主交换节点21上下叠置。例如,主交换单元20包括两个主交换节点21,两个主交换节点21叠置,如图5所示。在一些实施方式中,连接器30为正交连接器,可以是电连接器,也可以是光连接器。不难理解,连接器30的管脚数和信号速率可灵活扩展。
在本实施例中,冷却单元40与主交换单元20叠置,用于冷却刀片单元10和主交换单元20散发的热量,降低服务器的温度,以支持更高热耗的大功率芯片,降低服务器的PUE(电源使用效率)。
在一些实施方式中,冷却单元40包括风冷模块41,其中,风冷模块41与主交换单元20上下叠置设置,风冷模块41可以采用一组风扇或多组风扇。
在一些实施方式中,冷却单元40包括液冷模块42。液冷模块42包括进出水插头、分水器421、接水盘422、漏液检测器(图中未示出)和漏液导出器(图中未示出),接水盘422设置在分水器421的下方,漏液检测器设置在接水盘422的内侧,漏液导出器与接水盘422连通。分水器421将刀片单元10和主交换单元20产生的热量排出,接水盘422用于收集分水器421的漏液,漏液检测器用于检测分水器421是否存在漏液。当分水器421发生漏液时,接水盘422会聚集漏液,漏液检测器检测到漏液时将发出漏液提醒信息,同时可以启动漏液导出器,将接水盘422内的漏液排出,避免漏液损坏服务器内部的单板。
在一些实施方式中,进出水插头采用盲插式插头。如图4所示,盲插式进出水插头与进水管423和出水管424插接。进水管423和出水管424固定于服务器的机框的后部,并与内部的分水器421连接。盲插式插头不仅设计简洁,而且插拔维护方便。
在一些实施方式中,冷却单元40包括风冷模块41和液冷模块42,利用风冷模块41和液冷模块42对服务器进行散热,将刀片单元10、主交换单元20和电源产生的热量排出。风冷和液冷结合的方式可以提高散热效率,解决服务器散热的瓶颈,提高服务器的处理能力和交换带宽。
在一些实施方式中,可以将液冷模块42设置在服务器中功耗较高的处理器、芯片位置,利用液冷模块对高功耗的芯片和处理器进行散热,将风冷模块41设置在功耗相对较低的芯片、处理器位置,利用风冷模块41对功耗相对较低的芯片和处理器位置进行散热。
在一些实施方式中,如图5和图6所示,当冷却单元40包括风冷模块41,而不包括液冷模块42时,可以在液冷模块42的位置设置辅助交换单元70,以增加服务器的带宽。
在一些实施方式中,如图3所示,后插机构还包括电源50和机框管理单元60,电源50和机框管理单元60分别与前插机构电连接。其中,电源50用于提供电能,以满足服务器的正常运行。机框管理单元60用于负责收集管理框内部件的基本信息,单板上下电、各类传感器信息,风扇转速读取及调整,液体泄漏检测,电源监控,固件加载及更新操作等,使服务器高效合理的运行。
在一些实施方式中,电源50和机框管理单元60通过固定在背板80上的连接器分别与前插机构电连接。换言之,在背板80上固定有连接器,电源50与前插机构通过连接器电连接,例如,连接器的两个插接端一端插接电源50,另一端插接前插机构,实现电源50与前插机构的电连接。机框管理单元60与前插机构通过连接器电连接,例如,连接器的两个插接端一 端插接,机框管理单元60,另一端插接前插机构,实现机框管理单元60与前插机构的电连接。在一些实施方式中,电源50和机框管理单元60还可以通过固定在背板80上的连接器与其他模块或单元连接。其中,连接器可以根据连接对象选择不同功能的连接器,以实现相应的信号连接。
在一些实施方式中,电源50包括主电源和辅助电源,主电源为服务器提供电能;辅助电源为服务器提供备用电源,以提高电源50的可靠性。
在一些实施方式中,机框管理单元60包括主机框管理单元和辅助机框管理单元,辅助机框管理单元为主机框管理单元的备用管理单元。不难理解,主机框管理单元和辅助机框管理单元可以互为主备,只要机框管理单元60能够正常服务。
本实施例提供的服务器可用于超融合服务器,也可以用于其它需要高集成度、高处理能力的服务器。
本公开实施例提供的服务器,刀片单元和主交换单元通过连接器以正交方式连接,可以增加刀片单元和主交换单元之间接口带宽,提高传输效率,而且使得背板脱离了高速信号对背板材料的约束,增强了服务器的继承性和扩展性。
本领域普通技术人员可以理解,上文提及的功能模块/单元可以被实施为硬件及其适当的组合。在硬件实施方式中,在以上描述中提及的功能模块/单元之间的划分不一定对应于物理组件的划分;例如,一个物理组件可以具有多个功能,或者一个功能或步骤可以由若干物理组件合作执行。
在本公开中,虽然采用了具体术语,但它们仅用于并仅应当被解释为一般说明性含义,并且不用于限制的目的。在一些实施例中,对本领域技术人员显而易见的是,除非另外明确指出,否则可单独使用与特定实施例相结合描述的特征、特性和/或元素,或可与其它实施例相结合描述的特征、特性和/或元件组合使用。因此,本领域技术人员将理解,在不脱离由所附的权利要求阐明的本公开的范围的情况下,可进行各种形式和细节上的改变。

Claims (12)

  1. 一种服务器,包括:前插机构和后插机构,所述前插机构包括刀片单元,所述后插机构包括主交换单元,其中,所述刀片单元和所述主交换单元通过连接器以正交方式连接。
  2. 根据权利要求1所述的服务器,其中,所述连接器为正交连接器。
  3. 根据权利要求2所述的服务器,其中,所述正交连接器为电连接器或光连接器。
  4. 根据权利要求1-3任一项所述的服务器,其中,多个所述刀片单元沿第一方向布置成行,多个所述主交换单元沿与所述第一方向垂直的第二方向布置成列,将多个所述刀片单元和多个所述主交换单元相连接的多个所述连接器布置在与所述第一方向和所述第二方向均平行的同一平面中。
  5. 根据权利要求4所述的服务器,其中,每个所述刀片单元呈板状,并沿竖直平面延伸,每个所述主交换单元呈板状,并沿水平面延伸。
  6. 根据权利要求1-5任一项所述的服务器,其中,所述后插机构还包括冷却单元,所述冷却单元与所述主交换单元叠置。
  7. 根据权利要求6所述的服务器,其中,所述冷却单元包括风冷模块和/或液冷模块。
  8. 根据权利要求7所述的服务器,其中,所述液冷模块包括进出水插头、分水器、接水盘、漏液检测器和漏液导出器,所述进出水插头与所述分水器连通,所述接水盘设置在分水器的下方,所述漏液检测器设置在所述接水盘的内侧,所述漏液导出器与所述接水盘连通。
  9. 根据权利要求8所述的服务器,其中,所述进出水插头为盲插式进出水插头。
  10. 根据权利要求1-5任一项所述的服务器,其中,所述后插机构还包括电源和机框管理单元,所述电源和所述机框管理单元分别与所述前插机构电连接。
  11. 根据权利要求10所述的服务器,其中,所述电源包括主电源和辅助电源,所述主电源为服务器提供电能;所述辅助电源为所述服务器提供备用电源。
  12. 根据权利要求1至11任一项所述的服务器,其中,所述后插机构还包括辅助交换单元和背板,所述辅助交换单元与所述刀片单元固定于背板,并利用连接器信号连接。
PCT/CN2021/099082 2020-06-11 2021-06-09 服务器 WO2021249422A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202010531347.4 2020-06-11
CN202010531347.4A CN113805667A (zh) 2020-06-11 2020-06-11 服务器

Publications (1)

Publication Number Publication Date
WO2021249422A1 true WO2021249422A1 (zh) 2021-12-16

Family

ID=78845348

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/099082 WO2021249422A1 (zh) 2020-06-11 2021-06-09 服务器

Country Status (2)

Country Link
CN (1) CN113805667A (zh)
WO (1) WO2021249422A1 (zh)

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102544954A (zh) * 2010-11-11 2012-07-04 泰科电子公司 用于连接器系统的电缆组件
US20130178107A1 (en) * 2012-01-09 2013-07-11 Tyco Electronics Corporation Connector assembly for interconnecting electrical connectors having different orientations
CN104137659A (zh) * 2012-02-24 2014-11-05 泰科电子公司 用于互连通信系统中的卡模块的电缆组件
CN104137663A (zh) * 2012-02-22 2014-11-05 泰科电子公司 配置为在配合操作过程中对准通信连接器的连接器组件
CN108270097A (zh) * 2017-12-29 2018-07-10 曙光信息产业(北京)有限公司 一种正交高速背板
CN111273742A (zh) * 2019-12-04 2020-06-12 深圳市时代通信技术有限公司 一种基于正交构架的高密度服务模块化系统

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102544954A (zh) * 2010-11-11 2012-07-04 泰科电子公司 用于连接器系统的电缆组件
US20130178107A1 (en) * 2012-01-09 2013-07-11 Tyco Electronics Corporation Connector assembly for interconnecting electrical connectors having different orientations
CN104137663A (zh) * 2012-02-22 2014-11-05 泰科电子公司 配置为在配合操作过程中对准通信连接器的连接器组件
CN104137659A (zh) * 2012-02-24 2014-11-05 泰科电子公司 用于互连通信系统中的卡模块的电缆组件
CN108270097A (zh) * 2017-12-29 2018-07-10 曙光信息产业(北京)有限公司 一种正交高速背板
CN111273742A (zh) * 2019-12-04 2020-06-12 深圳市时代通信技术有限公司 一种基于正交构架的高密度服务模块化系统

Also Published As

Publication number Publication date
CN113805667A (zh) 2021-12-17

Similar Documents

Publication Publication Date Title
US6533587B1 (en) Circuit board riser
US7911793B2 (en) Case and rack system for liquid submersion cooling of electronic devices connected in an array
US7764511B2 (en) Multidirectional configurable architecture for multi-processor system
US9277680B2 (en) Blade computer system
US7978472B2 (en) Liquid-cooled cooling apparatus, electronics rack and methods of fabrication thereof
US6563706B1 (en) Typically high-availability information storage product
US20130077223A1 (en) Server
US10624228B2 (en) Rack mount case storage system separably coupled to body
CN109324989A (zh) 包括多个单独可移除子载体的标准形状因子电子模块载体
CN105230144A (zh) 半宽板的集中机箱架构
CN108874711B (zh) 一种优化散热的硬盘背板系统
TW201933138A (zh) 計算裝置及應用其之機架伺服器
WO2021249422A1 (zh) 服务器
CN104156037A (zh) 基于云服务的高密度微服务器
CN212623908U (zh) 一种新型服务器构架
US11317533B2 (en) Heat sink arrangements for data storage systems
CN111341359B (zh) 一种服务器高密度硬盘背板分板装置
CN218866444U (zh) 内存冷却组件、液冷模组及服务器
CN213876621U (zh) 一种vpx机箱
CN217587961U (zh) 一种基于双路国产cpu的人工智能服务器硬件架构
JP2013069087A (ja) 電子部品の実装構造
WO2024001749A1 (zh) 一种光模块的液冷结构及光模块
CN214670442U (zh) 一种飞腾八路服务器
WO2022062926A1 (zh) 一种多节点服务器、机柜式服务器以及刀片式服务器
CN113392061A (zh) 一种飞腾八路服务器及其设计方法

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21821593

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 24.05.2023)

122 Ep: pct application non-entry in european phase

Ref document number: 21821593

Country of ref document: EP

Kind code of ref document: A1