Bandwidth-Proportioned Datacenters

• US 20110258290A1
• Filed: 04/23/2010
• Published: 10/20/2011
• Est. Priority Date: 04/19/2010
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Hunt, Galen C., Nightingale, Edmund B., Elson, Jeremy E., Maltz, David A., Howell, Jonathan R.
• Granted Patent: US 8,438,244 B2
• Time in Patent Office: Days
• US Class Current: 709/219
• CPC Class Codes: G06F 16/182 Distributed file systems

Abstract

A system including at least one storage node and at least one computation node connected by a switch is described herein. Each storage node has one or more storage units and one or more network interface components, the collective bandwidths of the storage units and the network interface components being proportioned to one another to enable communication to and from other nodes at the collective bandwidth of the storage units. Each computation node has logic configured to make requests of storage nodes, an input/output bus, and one or more network interface components, the bandwidth of the bus and the collective bandwidths of the network interface components being proportioned to one another to enable communication to and from other nodes at the bandwidth of the input/output bus.

Claims

1. A system comprising:

at least one storage node having one or more network interface components and one or more storage units configured to store data, each network interface component and each storage unit having an associated bandwidth, the collective bandwidth of the network interface components and the collective bandwidth of the storage units being proportioned to one another to enable communication to and from other nodes at the collective bandwidth of the storage units;

at least one computation node having a processor, logic configured to read or write data to the at least one storage node, and one or more network interface components, each network interface component having an associated bandwidth; and

one or more switches connecting the at least one storage node to the at least one computation node such that a collective bandwidth available between the at least one computation node and the at least one storage node is greater than or equal to a smaller of either the collective bandwidth of the storage units on the at least one storage node or a collective bandwidth of the network interface components of the at least one computation node.

2. The system of claim 1, wherein the at least one storage node comprises a cluster of storage nodes and the at least one computation node comprises a cluster of computation nodes.

3. The system of claim 2, wherein the system is a datacenter.

4. The system of claim 1, wherein the at least one storage node and the at least one computation node are remote from one another.

5. The system of claim 4, wherein the proportioned bandwidth of the at least one storage node and of the at least one computation node enables the logic to read and write data to the at least one storage node with performance substantially equivalent to reading and writing data to a local storage device of the at least one computation node.

6. The system of claim 1, wherein the network interface components of the at least one storage node or the network interface components of the at least one computation node include at least one of a network interface card, a modem, or an optical interface.

7. The system of claim 1, wherein:

the at least one storage node comprises a plurality of storage nodes, each storage node being a server,

the at least one computation node is a client, and

the system includes a metadata server that maps data to storage nodes to enable the client to retrieve the mapped data.

8. The system of claim 1, wherein the one or more switches comprise a plurality of connected switches each having an incoming transmission bandwidth and an outgoing transmission bandwidth proportioned to one another.

9. The system of claim 8, wherein the one or more switches comprise a network that provides sufficient bandwidth between the compute nodes and the storage nodes for each of the at least one compute nodes to simultaneously read a tract from at least one of the storage nodes.

10. A storage node comprising:

one or more network interface components, each network interface component having a bandwidth;

a processor coupled to the network interface components and configured to process requests received from remote computation nodes; and

one or more storage units coupled to the processor, each storage unit having a bandwidth, the collective bandwidth of the storage units and the collective bandwidth of the network interface components being proportioned to one another to enable communication to and from other nodes at the collective bandwidth of the storage units.

11. The storage node of claim 10, wherein the collective bandwidth of the storage units and the collective bandwidth of the network interface components are within a predefined tolerance of one another.

12. The storage node of claim 11, wherein the collective bandwidth of the storage units matches the collective bandwidth of the network interface components.

13. The storage node of claim 10, wherein the storage units store data in a plurality of tracts, each tract belonging to a byte sequence comprised of multiple tracts, the multiple tracts of the byte sequence being distributed across multiple storage units.

14. The storage node of claim 10, wherein the storage node comprises a greater number of network interface components than storage units or a greater number of storage units than network interface components.

15. The storage node of claim 10, wherein one of the storage units is selected from a disk drive, a permanent storage drive, random access memory, an electrically erasable programmable read-only memory, a Flash Memory, a miniature hard drive, a memory card, a compact disc (CD), a digital versatile disk (DVD) an optical storage drive, a magnetic cassette, a magnetic tape, or a magnetic disk storage.

16. A computation node comprising:

a processor;

logic configured to be executed by the processor to formulate requests for remote storage nodes;

an input/output bus coupled to the processor, the input/output bus having a bandwidth; and

one or more network interface components, each network interface component having a bandwidth, the collective bandwidth of the network interface components and the bandwidth of the input/output bus being proportioned to one another to enable communication to and from other nodes at the bandwidth of the input/output bus.

17. The computation node of claim 16, wherein the bandwidth of the input/output bus and the collective bandwidth of the network interface components are within a predefined tolerance of one another.

18. The computation node of claim 17, wherein the bandwidth of the input/output bus matches the collective bandwidth of the network interface components.

19. The computation node of claim 16, wherein the requests include read and write requests.

20. The computation node of claim 19, wherein the write request is one of an atomic append or a random write and choice of whether to perform the write request as an atomic append or a random write is determined by whether byte sequence being written to is opened in an atomic append mode or a random write mode.