server rack

Abstract

A server rack that is efficient to operate and maintain is provided. A server rack housing a cooling device within a chassis includes: a first access port located on a first side of the server rack for housing at least one server; a second access port located on a second side opposite to the first access port; a bus located between the at least one server and the second access port within the chassis, and closest to the at least one server, for supplying power to the at least one server and carrying a high-voltage current; a power distribution unit (PDU) located closest to the bus and carrying a lower-voltage current than the bus; a manifold located closest to the second access port within the chassis and supplying refrigerant between the at least one server and the cooling device; and an opening located on the bottom surface of the chassis for supplying refrigerant to the cooling device via a pipe.

Description

Technical Field

[0001] This invention relates to a server rack. Prior Technology

[0002] Previously, a server rack has been disclosed in which a cooling device is disposed outside the server rack housing the server, and the refrigerant circulates from the cooling device to the server through pipes (see, for example, Patent Document 1). [Previous Technical Documents] [Patent Literature]

[0003] [Patent Document 1] Japanese Patent Application Publication No. 2011-86095 Summary of the Invention

[0004] [Solutions] A server rack according to one embodiment of the present invention houses a cooling device within a chassis and includes: a first access port located on a first side of the server rack for housing at least one server; a second access port located on a second side opposite to the first access port; a busbar located between the at least one server and the second access port within the chassis, and positioned closest to the at least one server to provide power to the at least one server by allowing a high-voltage current to flow through it; a power distribution unit (PDU) located between the busbar and the second access port, and positioned closest to the busbar, and allowing a lower-voltage current to flow through it; a manifold located closest to the second access port within the chassis, and supplying refrigerant between the at least one server and the cooling device; and an opening located on the bottom surface of the chassis, through which a pipe supplies refrigerant to the cooling device.

[0005] In a server rack related to one embodiment of the present invention, manifolds can be individually configured on both sides of the second working port, and each manifold is connected to the cooled refrigerant and the uncooled refrigerant.

[0006] In a server rack related to one embodiment of the present invention, the cooling device may be located at the bottom of the chassis.

[0007] In a server rack of one embodiment of the present invention, the server rack can house a plurality of cooling devices and a plurality of servers within a chassis; each manifold disposed on both sides of a second operating port is connected in a manner that transmits refrigerant between the respective cooling devices; the plurality of servers are connected in a manner that transmits refrigerant to any manifold.

[0008] In one embodiment of the present invention, the server rack is capable of mounting at least two types of servers: one conforming to the Electronic Industries Alliance (EIA) standard and the other conforming to the Open Computing Programming Protocol (OCP) standard. Simple Explanation of the Diagram

[0009] Figure 1 is a front perspective view of an example of a server rack related to the implementation configuration. Figure 2 is a rear perspective view of an example of a server rack related to the implementation configuration. Figure 3 shows the configuration of the server rack cooling system installed inside the chassis. Figure 4 is a schematic diagram of the server rack related to the implementation configuration. Figure 5 shows an example of a method in which a cooling device for a server rack is installed inside the server rack chassis, according to an implementation configuration. Figure 6 shows the rear bottom of the server rack when the cooling device is located at the very bottom of the server rack in the relevant implementation configuration. Implementation

[0010] Next, embodiments of the present invention will be described with reference to the accompanying drawings. In the drawings relating to the embodiments, identical or similar parts are given the same or similar symbols. Furthermore, it is undeniable that the drawings also include parts with different relationships to each other.

[0011] Furthermore, the embodiments described herein are apparatuses or methods used to embody the technical concept of the present invention, and the technical concept of the present invention does not limit the arrangement of the constituent elements to the following. The technical concept of the present invention can be modified in various ways within the technical scope defined by the claims described in the patent claims.

[0012] In recent years, with the advancement of artificial intelligence (AI) technology, the demand for server computing power has been increasing rapidly, and the design and development of data centers in various regions are constantly progressing.

[0013] It is known that as server computing power increases, so does the heat generated. The increased computing power demands accompanying the development of artificial intelligence technology can be considered equivalent to an increase in server heat generation. Therefore, recent data center design and development need to address this increase in server heat generation. To address this increase in heat generation, in addition to existing cooling technologies, new cooling systems must be developed and provided.

[0014] However, the server cooling system exists independently of the server and server rack. Since servers in data centers generate a large amount of heat, it is desirable for the server, server-related components, and cooling system to be tightly integrated and optimized within the rack. However, in currently available server racks, the cooling system is an added-on design; the cooling piping and cooling units are not assembled into the server rack, leading to complex cooling system setup issues. Furthermore, in server racks with added cooling systems, the piping is exposed, complicating operations such as server replacement, increasing the workload due to refrigerant leaks, and creating a complex work environment – ​​a clear lack of consideration for maintainability.

[0015] To address these issues, according to one embodiment of the present invention, a server rack is provided that mounts a cooling device inside a chassis and has a first working port for server setup and other operations, and a second working port for power supply and refrigerant supply to the server. By including: a busbar located between at least one server and the second working port within the chassis, and closest to at least one server; a power distribution unit (PDU) located closest to the busbar; a manifold located closest to the second working port within the chassis; and an opening located on the bottom surface of the chassis, through which refrigerant is supplied to the cooling device, this provides a server rack that is more efficient in operation and considers maintainability compared to existing server racks where the cooling device is located outside the server rack and the refrigerant circulates from the cooling device to the server through a pipe.

[0016] <Structure> Figures 1 and 2 are perspective views of the front (first side) 19 and rear (second side) 20 of an example of a server rack 10 related to this embodiment. The server rack 10 shown in Figures 1 and 2 has its outer panel removed, and the components constituting the server rack 10 can be seen from the outside. However, in reality, the server rack 10 is surrounded by an outer panel.

[0017] The server rack 10 shown in Figures 1 and 2 consists of a first access port 11, a cooling device 12, a bus 13, a power distribution unit (PDU) 14, a manifold 15, and a second access port 18, all located within the chassis.

[0018] The first access port 11 is located at the front side 19 of the server rack 10. Figure 1 shows the state with at least one server 17 already installed. At least one server 17 is installed inside the chassis of the server rack 10 through the first access port 11. The first access port 11 can also be closed by a door panel (not shown) except when installing or maintaining at least one server 17. In addition, the width of the first access port 11 and the size of the server rack 10 can be used to accommodate the 21-inch server (Open Computing Program (OCP) standard) which is currently moving towards standardization and the existing 19-inch server (Electronic Industries Alliance (EIA) standard).

[0019] The second access port 18 is an access port located on the rear side 20 of the server rack 10, which is opposite to the front side 19. In addition to maintaining the internal components of the server rack 10, the second access port 18 can also be closed by means of a door panel or the like (not shown).

[0020] Bus Bar 13 is a plate-shaped conductor. Bus Bar 13 carries high-voltage current and supplies power to at least one server 17. When at least one server 17 is installed in the server rack 10 chassis, Bus Bar 13 is positioned closest to at least one server 17. When at least one server 17 is pushed from the first port 11 into the position of Bus Bar 13 in the server rack 10 chassis, at least one server 17 will be connected to Bus Bar 13, and power will be supplied from Bus Bar 13 to at least one server 17.

[0021] The power distribution unit 14 carries a lower voltage current than the bus and provides power to network devices, except for at least one server 17. The power distribution unit 14 is located between the bus and the second port in the chassis of the server rack 10, and is closest to the bus.

[0022] Manifold 15 is a tubular component that holds the refrigerant for cooling at least one server 17 inside and circulates the refrigerant between the at least one server 17 and the cooling device 12. As shown in Figures 1 and 2, manifold 15 is a plate-shaped component, but in reality, tubular components are respectively provided inside the plate-shaped component for circulating the refrigerant at a higher temperature after cooling (the refrigerant before cooling), and for circulating the refrigerant at a lower temperature before cooling (the refrigerant after cooling). Manifold 15 is located between the power distribution unit 14 and the second operating port 18, and is closest to the second operating port 18. As shown in Figures 1 and 2, manifolds can be individually arranged on both sides of the second operating port, and each manifold connects both the cooled refrigerant and the refrigerant before cooling.

[0023] The cooling device 12 is a cooling device installed inside the chassis of the server rack 10. The cooling device 12 cools the refrigerant circulating in the manifold 15. The cooling device 12 can be installed anywhere inside the chassis of the server rack 10. As an example, in this embodiment, the cooling device 12 is installed at the bottom of the chassis of the server rack 10.

[0024] Figure 3 shows a rear view of the server rack 10. The manifold 15 is arranged along the side wall of the server rack 10 chassis on both sides of the second access port 18. The power distribution unit 14, like the manifold 15, is arranged along the side wall and is located behind the manifold 15 in Figure 3, but is not shown. The bus 13 is not along the side wall of the server rack 10 chassis; in Figure 3, it is located in the center of the server rack 10.

[0025] Figure 4 is a schematic diagram showing the server rack 10. As shown in Figures 1, 2, and 4, in the server rack 10 of this embodiment, the manifold 15 is located closest to the second working port 18, the power distribution unit 14 is adjacent to the manifold 15 and located after the manifold 15, closest to the second working port 18, and the bus 13 is located furthest from the second working port 18 compared to the manifold 15 and the power distribution unit 14. The bus 13 carries a high voltage current, and there is a risk of electric shock if an operator touches the bus 13 during maintenance of the server rack 10. In the server rack 10 of this embodiment, the bus 13 is located furthest from the second working port 18, thereby reducing the risk of electric shock due to contact with the bus 13 during maintenance of the power distribution unit 14 and the manifold 15 at the second working port 18.

[0026] Furthermore, the manifold 15 and power distribution unit 14 are arranged along the side wall of the server rack 10 chassis, but the bus 13 is not arranged along the side wall of the server rack 10 chassis, but is arranged in the center of the server rack 10. While the power distribution unit 14 and manifold 15 are arranged along the side wall of the server rack 10 chassis, the bus 13 is arranged in the center of the server rack 10. Since the bus 13 is arranged in a position not adjacent to either the power distribution unit 14 or the manifold 15, this also reduces the risk of electric shock due to contact with the bus 13.

[0027] Figure 5 shows an example of a method for installing the cooling device 12 inside the chassis of the server rack 10. Figure 3 shows the cooling device 12 installed at the bottom of the chassis from outside the server rack 10. By installing the cooling device 12 at the bottom of the server rack 10 chassis, even if the refrigerant circulating from the cooling device 12 leaks due to damage or other reasons, the server can be prevented from being damaged by the refrigerant. Furthermore, the refrigerant used in the server rack 10 of this embodiment has insulating properties, which can prevent serious damage to the server even in the event of refrigerant leakage.

[0028] Generally, cooling systems are designed to cool servers to a certain temperature. This certain temperature is expressed, for example, by a parameter such as Thermal Design Power (TDP). In other words, the number of servers a cooling system can cool depends on the heat generated by those servers. If the total heat generated by the servers connected to a cooling system exceeds the cooling capacity of the system, the servers may malfunction.

[0029] Although only one cooling device 12 is shown in Figure 5, multiple cooling devices may be installed in the chassis of the server rack 10, for example, to cope with the heat generated by the server.

[0030] This improves the overall cooling capacity of the chassis, and increases the number of servers that can be cooled, thereby improving computing power.

[0031] Furthermore, multiple cooling units can be connected to different manifolds. Servers can also be connected to cooling units via different manifolds. That is, servers mounted in the same chassis are not limited to all being connected to the same cooling unit. Specifically, for example, when two cooling units (cooling unit A and cooling unit B) and four servers (server α, server β, server γ, and server δ) are mounted in the same chassis, servers α and β can be connected to cooling unit A, while servers γ and δ can be connected to cooling unit B. Additionally, the connection between cooling units and servers is not limited to the above example; three of the four servers can also be connected to the same cooling unit. The number of servers connected to a single cooling unit can be determined, for example, based on the server's designed heat output or the server's heat output under normal operating conditions.

[0032] Figure 6 shows the rear bottom of the server rack 10 with the cooling units 12 installed at the very bottom of the server rack 10's chassis. In Figure 6, as an example, two cooling units 12 are installed within the server rack 10's chassis. An opening 61 is provided on the bottom surface of the server rack 10's chassis for supplying refrigerant to the cooling units 12 via pipes. The two cooling units 12 are respectively connected to: a pipe 63 for circulating the refrigerant at a higher temperature after the target to be cooled (used to cool at least one server 17); a pipe 64 for circulating the refrigerant at a lower temperature before the target to be cooled; and pipes 62 and 65 for circulating the refrigerant used to cool the cooling unit 12 itself. The pipe 63 for circulating the higher-temperature refrigerant and the pipe 64 for circulating the lower-temperature refrigerant are respectively connected to a manifold 15. Pipes 62 and 65, which circulate refrigerant to cool the main body of the cooling device 12, are connected to the outside of the server rack 10 chassis via openings 61 to supply refrigerant.

[0033] The refrigerant circulating within pipes 62 and 65 to cool the cooling device 12 can be, for example, water. Pipes 62 and 65 are connected via openings 61 to a water supply device (not shown) outside the server rack 10 chassis. In the prior art, pipes used to circulate refrigerant for cooling equipment installed inside the server rack 10 chassis are routed from the side of the chassis to the outside, rather than from the bottom of the server rack 10 chassis. Therefore, pipes routed from the side of the chassis to the outside would hinder maintenance of the server or server rack, and there is also a risk of damage to the pipes due to contact with operators or other components. Especially when at least one of the refrigerants used to cool the server 17 or to cool the cooling device 12 is conductive, if a refrigerant leak occurs due to damage to the piping, etc., other devices within the server rack that are physically located below the leak location will be exposed to the conductive refrigerant, increasing the risk of malfunction. In the server rack 10 of this embodiment, the refrigerant supply equipment located outside the server rack 10 chassis is positioned directly below the server rack 10. The refrigerant supply equipment is connected to the server rack 10 directly below via pipes 62 and 65 through openings 61. This avoids obstruction of server or server rack maintenance and pipe damage caused by pipes extending outside the chassis. Furthermore, as shown in Figures 2 and 6, a bracket 16 can be provided at the bottom of the server rack 10 chassis to properly connect the pipes 62 and 65 to the refrigerant equipment outside the chassis. This configuration is particularly effective when using a conductive refrigerant for cooling the server 17, or at least one of the refrigerants used to cool the body of the cooling device 12.

[0034] As an example, the cooling method for at least one server 17 used in the server rack 10 related to this embodiment can be a direct liquid cooling method. Direct liquid cooling systems include: single-phase direct liquid cooling, where either the refrigerant at a higher temperature after cooling or the refrigerant at a lower temperature before cooling is a liquid; and two-phase direct liquid cooling, where the refrigerant at a higher temperature after cooling is a liquid, and the refrigerant at a lower temperature before cooling is a gas. Two-phase direct liquid cooling is a method that uses a refrigerant with a boiling point of approximately 30°C as the refrigerant, circulating this refrigerant within a metal plate, called a cold plate, which is in contact with the object to be cooled. The liquid refrigerant in the cold plate evaporates due to heating by the object to be cooled, and the heat of vaporization cools the cold plate, thereby cooling the object to be cooled. The refrigerant that becomes gaseous in the cold plate is cooled and / or compressed by a cooling device to return to a liquid state, and then circulated back from the cooling device to the cold plate.

[0035] In existing air-cooled server racks, the implementation of two-phase direct liquid cooling equipment was not anticipated. Furthermore, because the piping for power supplies or external refrigerant equipment cannot be accommodated within the server rack, the piping is sometimes exposed. Therefore, personnel operating the servers, staff working on nearby servers, or robots moving within the server room may accidentally damage the piping. Thus, it is necessary to redesign the external refrigerant equipment or its operation. Regarding the server rack 10 of this embodiment, all the necessary components for a two-phase direct liquid cooling system are housed within the server rack, eliminating the need for complex wiring or scheduling management. This allows for the implementation of a two-phase direct liquid cooling system, thereby improving operational efficiency. Moreover, by implementing the server rack 10 of this embodiment, the equipment conversion can be quickly completed even when replacing air-cooled equipment with two-phase direct liquid cooling equipment or facilities.

[0036] Furthermore, in the past, the power supply and cooling equipment for servers were often separate entities, or located in slots within the server rack. During server deployment and maintenance, this required managing multiple independent devices, or separately managing the front and back of the server rack. In the server rack 10 of this embodiment, all the necessary components for operating the power supply or the two-phase direct liquid cooling system are concentrated at the rear of the server rack, i.e., the back. Since there is no need to separately manage and maintain these components, operational efficiency is improved. The power supply, wiring, or piping for the two-phase direct liquid cooling system is located in front of the busbar supplying power to the server, thus reducing the risk of operator contact with the high voltage of the busbar and subsequent electric shock.

[0037] Furthermore, existing technologies, in order to accommodate server racks conforming to server specifications, have not even considered the rack depth required for piping in two-phase direct liquid cooling systems. When a self-air-cooled server is replaced with a two-phase direct liquid cooling system, the wiring or piping cannot be accommodated within the rack, resulting in facility redesign, more complex operator operations, and risks of accidental contact. However, in this invention, the piping can be completely accommodated within the rack, and the piping can be easily connected from the bottom of the rack, thus improving the efficiency of installation and operation.

[0038] As described above, the server rack associated with this embodiment has both versatility and scalability, allowing for the high-density configuration of at least one server. Furthermore, by arranging the power distribution unit and manifold at the rear of the chassis, safe and robot-friendly operation is feasible.

[0039] As stated above, it is undeniable that the present invention includes various embodiments not described herein. Therefore, the technical scope of the present invention is determined solely by referring to the specific inventive aspects related to the appropriate patent application claims from the foregoing description.

[0040] This invention aims to provide a server rack that is both highly efficient and maintainable. In recent years, with the widespread adoption of Artificial Intelligence (AI) and its further development towards Artificial General Intelligence (AGI) and Artificial Super Intelligence (ASI), computing resources have expanded significantly. By using this invention, it is possible to achieve both the versatility and scalability of high-density server configuration, while also ensuring operational safety and robot-friendly operation, thereby contributing to SDG 9: "The foundation for industrial and technological innovation."

[0041] 10: Server rack 11: First work port 12: Cooling device 13: Busbar 14: Power Distribution Unit (PDU) 15: Manifold 16: Bracket 17: At least one server 18: Second work port 19: Front 20: Rear side 61: Opening 62, 63, 64, 65, 71, 72, 74, 75: Pipe body 70: Cold Plate 73: Intermediate components

Claims

1. A server rack, comprising a server rack housing a cooling device within a chassis, comprising: a first access port disposed on a first side of the server rack for housing at least one server; a second access port disposed on a second side opposite to the first access port; a bus disposed between the at least one server and the second access port within the chassis, and positioned closest to the at least one server to provide power to the at least one server and allow high-voltage current to flow through it; a power distribution unit (PDU) disposed between the bus and the second access port, and positioned closest to the bus, and allowing low-voltage current to flow through it; a manifold disposed closest to the second access port within the chassis and supplying refrigerant between the at least one server and the cooling device; and an opening located on the bottom surface of the chassis and supplying refrigerant to the cooling device via a pipe.

2. The server rack as described in claim 1, wherein, Each of the aforementioned manifolds is individually configured on both sides of the aforementioned second working port, and each manifold is connected to the cooled refrigerant and the uncooled refrigerant.

3. The server rack as described in claim 1, wherein, The aforementioned cooling device is located at the very bottom of the aforementioned chassis.

4. The server rack as described in claim 2, wherein, The aforementioned server rack houses a plurality of cooling devices and a plurality of servers within a chassis; each of the aforementioned manifolds, located on either side of the aforementioned second operating port, is connected in a manner that transmits refrigerant between the respective cooling devices; the aforementioned plurality of servers are connected in a manner that transmits refrigerant to any of the aforementioned manifolds.

5. The server rack described in claim 1 is capable of hosting at least two servers: one conforming to the Electronic Industries Alliance (EIA) standard and the other conforming to the Open Computing Programming Protocol (OCP) standard.

Images