WO2016018398A1

WO2016018398A1 - Redundant power supply units

Info

Publication number: WO2016018398A1
Application number: PCT/US2014/049261
Authority: WO
Inventors: Thomas Lee Ellison; Herbert J. Tanzer
Original assignee: Hewlett-Packard Development Company, L.P.
Priority date: 2014-07-31
Filing date: 2014-07-31
Publication date: 2016-02-04

Abstract

A system of redundant power supply units including a node of a storage system and a redundant power supply unit (RPSU) conductively connected to the node. The RPSU may be powered completely off unless it is powering the node. The system includes a backup battery conductively connected to the RPSU and to the node. The backup battery may be charged by the RPSU and may provide power to the node. The backup battery may provide power to the node if the RPSU does not provide adequate power to the node until another RPSU turns on and adequately powers the node.

Description

REDUNDANT POWER SUPPLY UNITS

BACKGROUND

[0001] Organizations, consumers, and businesses often require the storage of digital information in large quantities. Storing this digital information can be done in many ways including storage area networks (SANs) or network- attached storage (NAS) among others. The services and products involved in storing digital information for these groups may generally be referred to as enterprise storage systems. Enterprise storage systems can include primary systems where data is directly accessed and my also include backup storage systems to house copies. Enterprise storage systems may be onsite or may be off-site and run by a third party and referred to as cloud storage. Every enterprise storage system requires, at some level, powered storage hardware. In order to provide greater data availability and more robust systems, many enterprise storage systems have redundant power sources and backup batteries all connected to aid in powering the powered storage hardware. However, in order to reduce costs of high power consumption, wear and tear to power supplying components, and in order to comply with relevant regulations, more efficient systems and methods are needed for providing power to storage systems from redundant power supplies.

BRIEF DESCRIPTION OF THE DRAWINGS

[0002] Certain examples are described in the following detailed description and in reference to the drawings, in which:

[0003] Fig. 1 is a block diagram of an example storage system that includes redundant power supply units;

[0004] Fig. 2 is a block diagram of an example system including several redundant power supply units configured to provide power to a node;

[0005] Fig. 3 is a flow diagram of an example process for managing a system of redundant power supply units providing power to a node; and

[0006] Figs. 4A, 4B, and 4C are schematic diagrams of a progression of RPSUs maintaining power to a node during the failure of an RPSU. DETAILED DESCRIPTION

The presence of multiple power supply units provides protection for a storage system from unexpected power loss issues. The power loss issues of a storage system may vary by the specific hardware and function in each node of the storage supply system. In some cases, power loss to a node containing hard drive disks may result in an interruption of data availability for an organization or client requiring data from the disks in that node. In other cases, power loss to a node containing a memory controller which stores data in volatile memory for later computation and storage may result in complete loss of that data. Accordingly, redundancy in power supply units may help protect against these and other power loss issues.

In many systems with redundant power supply units (RPSUs), power draw is divided between each of the RPSUs, thereby reducing the load on each individual power supply unit. However, as power supplies tend to operate most efficiently at peak loads, the equal division of a load between power supply units sacrifices overall efficient operation for redundancy. In other types of systems with redundant power supplies, the power load is not divided equally, but unused power supply units still remain powered on and in a low power or idle state. When in this idle state each power supply unit still represents a major contributor to the power consumption of the storage system.

[0007] Examples described herein disclose using RPSUs to power a node of a storage system in a phase-in/phase-out method to increase efficiency. The phase- in/phase-out method may be used to eliminate the idle power consumption of one power supply and operate another at closer to maximum capacity. The redundant hardware and method described herein also ensures that the phase-in/phase-out process does not become a "single point failure" source.

[0008] Fig. 1 is a block diagram of an example storage system that includes RPSUs. The example storage system 100 includes a conductive connection to mains power 102. Mains power 102 may refer to a general-purpose alternating- current electric power supply such as a power grid. A power distribution unit (PDU) 104 may couple the mains power 1 02 to a rack 106. A PDU 102 may also be referred to as a mains distribution unit, and includes a device receiving current and distributing it through multiple outputs. Examples of a power distribution unit include a power strip, a stand-alone cabinet unit that include main breakers, an internal bus bar for neutral and ground conductors, individual circuit breakers, or power monitoring panels, among others.

[0009] A rack 106 may be any of a number of different types of frames or enclosures used for mounting equipment. In some examples, the rack 106 may be a server rack. The storage system 100 in this example also includes a number of nodes 108-1 12 mounted on the rack 106. A first node 108 may contain a memory controller module. In this example, a memory controller module may control the placement and retrieval of data in a single node 108 or across several nodes that are communicatively connected. The first node 108 may also include volatile memory used to store data temporarily before supplying it to a processor or storing it in nonvolatile memory. The first node 108 may also contain nonvolatile memory.

Additional nodes 1 10-1 12 may contain memory and storage modules in the form of volatile memory for temporary data storage or nonvolatile memory for longer term storage. These memory and storage modules may include hard disk drives, Solidstate Storage Devices (SSDs), random access memory modules, read only memory modules, static random access memory (SRAM), dynamic random access memory (DRAM) and any other module which can hold data. A communicative connection may exist between the nodes 108-1 12.

[0010] In this example, each of the nodes 1 08-1 12 have RPSUs 1 14-1 18. As shown in the example block figure of a storage system 100, the RPSUs 1 14-1 1 8 are conductively connected to the PDU 1 04 and to each of the nodes 108-1 1 2. In this example, the RPSUs 1 14-1 18 are providing power to each of the nodes 108-1 12. In Fig. 1 , power is provided from the PDU 1 04 to the RPSUs 1 14-1 18 to the nodes 1 08- 1 12, through conductive connector 120.

[0011] The conductive connector 120 allows an electric current to flow between the RPSUs 1 14-1 18 in different nodes 1 08-1 12. In this example, the conductive connector 120 is seen splitting between the PDU 104 and each of the RPSUs 1 14- 1 18. Although the conductive connector 1 20 may be a pair of electrically conductive wires, any means of conducting electric current is included in this example. For example, the RPSUs 1 14-1 18 may couple to power rails built into the rack 106. Other systems may be used as well, such as transformer coupling between coils built into the rack 106 and coils built into each of the RPSUs 108-1 12. [0012] In other examples, the system of Fig. 1 may have the RPSUs 1 14-1 18 built in to the rack 106, built in to the nodes 1 08-1 12, or separately attachable to either. The PDU 104 may also provide power to the rack 106, which provides power to each of the RPSUs 1 14-1 1 8. The first node 1 08 is not limited to including a memory controller but may include the same memory and storage modules discussed for nodes 1 10-1 12.

[0013] Fig. 2 is a block diagram of an example system 200 including several redundant power supply units configured to provide power to a node. The like numbered items are as described with respect to Fig. 1 . In the system 200, three RPSUs 202, 206, and 21 0 are used to power a node 108. As described with respect to Fig. 1 , the node 108 may be mounted in a rack 106.

[0014] The conductive connection 120 splits in order to provide power to each RPSU 202, 206, and 210. If other nodes were also being supplied power by the PDU 104, additional splits in the conductive connection 120 may be present. Each RPSU 202, 206, and 210, for example RPSU 202, is also conductively connected its own backup battery 204 and the node 108. Each of the backup batteries 204, 208, and 21 2 are conductively connected to the node 108. Each of the connections to the node 1 08 by either a RPSU 202 or a backup battery 204 allows power to be provided to the node.

[0015] Although the system 200 has three RPSUs 202, 206, and 210, the system is not limited to three power supply units. In other examples, systems may include two RPSUs, three RPSUs, or more. Further, the PDU 1 04 may also be directly conductively connected to the backup batteries 204, 208, and 212. The RPSUs 202, 206, 210 may also be powered separately with the conductive connector 120 splitting again at each of the RPSUs 202, 206, 210 to separately provide electric current to each.

Fig. 3 is a flow diagram of an example process for managing a system of redundant power supply units providing power to a node. The process 300 begins at block 302, with a storage system powering a node of a storage system with a redundant power supply unit (RPSU) from a number of RPSUs, each conductively connected to the node. As only a single RPSU is being used to provide power, the operational RPSU may be operating at a more efficient load. At block 304, each backup battery that is conductively connected to both the node of the storage system and an RPSU is charged. In this example method, each backup battery may be conductively connected to the node to ensure that during a transition between individual RPSUs, the node is still receiving power.

At block 306, a backup battery powers the node when the RPSU powering the node fails. The backup batteries may power the node until another RPSU is activated to provide power to the node or, in case of complete power failure to allow the software to migrate all data to non-volitale storage prior to the batteries becoming fully discharged.

At block 308, another RPSU connected to the node is activated. This second RPSU may be powered on with power from its own backup battery, and may provide power to the node of the storage system. In other examples, the powered down RPSU may be powered on by power directly from the PDU or from a backup battery on another RPSU.

[0016] Figs. 4A, 4B, and 4C are schematic diagrams of an example of a progression of RPSUs maintaining power to a node during the failure of an RPSU. The like numbered items are as described with respect to Figs. 1 and 2. In Figs. 4A, 4B, and 4C, a PDU 1 04 provides power to each of the RPSUs 402 and 406 located inside a powered node 108. Each RPSU 402 and 406 is conductively connected to a backup battery 404 and 408 and the node 108. As noted above, this configuration is merely an example and in other examples the power may be provided to each backup battery 404 by the PDU 104 directly.

[0017] In Fig. 4A, RPSU 406 is providing power to the node 1 08, while RPSU 402 is powered completely off. Using a single RPSU to power the node saves energy and improves the efficiency of the system 400 over systems in which both power supplies are left running at low levels. In this example, the backup battery 404 coupled to RPSU 402 continues to charge, for example, from power provided by the conductive connection 120 from the PDU 104. If the backup battery 404 is completely charged, charging current is reduced to a level that maintains the battery charge. In this example, the power may be passed through charging circuitry in the generally unpowered RPSU 402. In other examples, the backup battery 404 may be connected and charged directly from a conductive connection to the PDU 104 or may be charged from the other RPSU 406. [0018] In Fig. 4B, RPSU 406 has failed, or been deactivated, and is no longer providing power. One or both of the backup batteries 404 and 408 may then provide power to the node 108. As the backup batteries 404 and 408 power the node 108, RPSU 402 may be powered on using either the power from the backup batteries 404 and 408, power from the PDU 104, or both.

[0019] In Fig. 4C, RPSU 402 is powered and providing power to the node 108. Each of the backup batteries 404 and 408 have stopped discharging to power the node 1 08 and have returned to drawing charge from the PDU 104 via their respective conductive connections. Throughout the transition from Fig. 4A to 4C, the node 1 08 remains powered. The transition from being powered by one RPSU to another is performed after the failure of one of the RPSUs 402 or 406. However, in other examples, the system may switch between the RPSUs 402 or 406 to decrease the wear on an individual RPSU. While in this example only two RPSUs are shown, this switching is also contemplated among three or more power supply units, so long as the RPSUs in operation may operate at near peak energy efficiency.

While the disclosed subject matter has been described with reference to illustrative examples, this description is not intended to be construed in a limiting sense.

Various modifications of the illustrative examples, as well as other examples of the subject matter, which are apparent to persons skilled in the art to which the disclosed subject matter pertains are deemed to lie within the scope of the disclosed subject matter.

Claims

CLAIMS What is claimed is:

1 . A system of redundant power supply units comprising:

a node of a storage system;

a redundant power supply unit (RPSU) conductively connected to the node wherein the RPSU is powered completely off unless it is powering the node; and

a backup battery conductively connected to the RPSU and to the node;

wherein the backup battery is charged by the RPSU and provides power to the node if the RPSU does not provide adequate power to the node until another RPSU turns on and adequately powers the node.

2. The system of claim 1 wherein the RPSU is most power efficient at a high percentage of its power output capacity.

3. The system of claim 2 comprising logic in a memory region of the RPSU that when executed enables another RPSU to alternatively turn on and power the node or not so the over-all power efficiency of the system of redundant RPSUs providing power to the node is maximized.

4. The system of claim 1 wherein the backup battery is located inside the

RPSU.

5. The system of claim 1 wherein the node of the storage system is either a memory controller or a plurality of hard drive disks.

6. The system of claim 1 wherein if the RPSU is not providing power to the node the backup battery is still being charged.

7. A method of managing redundant power supply units of a storage system comprising: powering a node of a storage system with a redundant power supply unit

(RPSU) wherein both the node and the power supply are conductively connected ^*to a backup battery;

charging the backup battery;

powering the node with the backup battery when the node is not receiving adequate power; and

activating another RPSU to power the node when the first RPSU fails to

provide adequate power to the node, wherein the second RPSU is completely unpowered until it is activated.

8. The method of claim 7 comprising powering a node that has data stored in volatile memory.

9. The method of claim 8 comprising copying the data of the node stored in volatile memory to a nonvolatile memory location when the first RPSU fails to provide adequate power to the node.

10. The method of claim 7 comprising alternately activating and

deactivating an RPSU so the over-all power efficiency of the system of is maximized.

1 1 . A non-transitory machine readable storage medium comprising instructions to direct a processor to manage the power consumption of a storage system with a system of redundant power supplies, the instructions to direct the processor to:

power a node of a storage system with a redundant power supply unit (RPSU) wherein the RPSU is coupled to a backup battery and the node of the storage system;

charge the backup battery wherein each backup battery is coupled to the

node of the storage system;

power the node with the backup battery when the PSU fails to fully power the node;

activate another RPSU to power the node wherein any RPSU not being used to power the node is completely unpowered; and stop powering the node with the backup battery when the second RPSU is fully powering the node.

12. The non-transitory machine readable storage medium of claim 1 1 comprising instructions that direct a processor to power a node that has data stored in volatile memory.

13. The non-transitory machine readable storage medium of claim 12 comprising instructions that direct a processor to move data of the node stored in volatile memory to a nonvolatile memory location when an RPSU that provides power to the node beings to fail.

14. The non-transitory machine readable storage medium of claim 12 comprising instructions that direct a processor to activate or deactivate an RPSU as needed such that the over-all power efficiency of the system is maximized.

15. The non-transitory machine readable storage medium of claim 1 1 comprising instructions that direct a processor to switch, at a set interval, the RPSU that provides power to the node such that the wear of providing power to the node is evenly spread between each of a plurality of RPSUs coupled to the node of the storage system.