Computing system with unified storage, processing, and network switch fabrics and method for making and using the same
A system and method for making and using a computing system with unified storage, processing, and network switch fabrics are provided. Processing nodes, either physical or virtual, are associated with intra-module ports, inter-module ports, and local storage spaces. A plurality of processing nodes are linked through intra-module ports to form processing modules. A plurality of the processing modules are further connected through inter-module ports to form the computing system. Several inter-module connection schemes are described, each of which can be adapted to use with existing network packet routing algorithms. Each processing node need only to keep track of the states of its directly connected neighbors, obviating the need for a high-speed connection to the rest processing nodes within the system. As a result, dedicated network switching equipment is not needed and network capacity grows naturally as processing nodes are added.
1. A computing system with a unified storage, processing and network switching fabrics, comprising:
a plurality of processing nodes;
a plurality of intra-module ports, each of the intra-module ports is associated with one of the processing nodes;
a plurality of inter-module ports; each of the inter-module ports is associated with one of the processing nodes;
a plurality of processing modules each comprising at least one of the processing nodes, each processing node is connected to all remaining ones of the processing nodes within the module in which that processing node is comprised, to form an intra-module network; and
an inter-module network comprising at least one connection between one of the inter-module ports on one of the processing modules to another one of the inter-module ports on another one of the processing modules, wherein the inter-module network comprises connections provided according to the method of connecting an inter-module port P on the processing modules (2 P *N+1) through (2 P *N+2 P-1 ) to a further inter-module port P on the processing modules (2 P *N+2 P-1 +1) through (2 P *N+ 2 P ), respectively, where N is [0, 1, . . . , (M/2 P ) −1], and P is [1, 2, . . . , P ], wherein the number of the processing modules is M and each of the processing modules has at least P inter-module ports.
2. A system according to claim 1 , wherein the intra-module connection network comprises at least one of:
a bus, a ring, a star, a mesh, and a crossbar switch.
3. A system according to claim 1 , wherein the processing nodes each comprise at least one of:
a physical node and a virtual node.
4. A system according to claim 1 , wherein the processing nodes each comprise at least one of:
a processing element, a memory controller, a memory, a storage controller, a storage device, and interfaces to intra-module and inter-module ports.
5. A system according to claim 1 , further comprising:
a further inter-module network comprising a series of linearly connected processing modules with the beginning and the end of the series further linked to each other to form a ring, using two inter-module ports from each of the connected processing modules.
6. A system according to claim 5 , further comprising:
a connection between two non-adjacent processing modules on the ring, via one inter-module port located on one of the two processing modules and not used for the linear connection, to another inter-module port located on the other of the two non-adjacent processing modules and not used for the linear connection.
7. A system according to claim 5 , further comprises at least one of:
a connection between a processing module N on the ring to processing a module N+S on the ring, by connecting one unused inter-module port on the processing module N to another unused inter-module port on the processing module N+S, wherein M is the number of the processing modules on the ring, S is an integer divisor of M, N is [x, x+S, . . . , x+(M−S), increment by S], and x is [1, 2, . . . , S−1];
a connection between a processing module N on the ring to processing a module N+S−1 on the ring, by connecting one unused inter-module port on the processing module N to another unused inter-module port on the processing module N+S−1, wherein M is the number of the processing modules on the ring, S is an integer divisor of M, N is [x, x+S, . . . , x+(M−S), increment by S], x is [1, 2, . . . , S−1]; and
a connection between a processing module N on the ring to a processing module N+S−r on the ring, by connecting one unused inter-module port on the processing module N to another unused inter-module port on the processing module N+S−r, wherein M is the number of the processing modules on the ring, S is an integer divisor of M, N is [x, x+S, . . . , x+(M−S), increment by S], x is [1, 2, . . . , S−1], and r is [2, 3, . . . , S−1].
8. A system according to claim 1 , further comprising:
a transmission module to transfer data packet from one of the processing nodes to another one of the processing nodes through the established inter-module connection.
9. A system according to claim 8 , further comprising:
a monitoring module measuring the data packet traffic pattern; and
a modification module to amend the established inter-module connection based upon the packet traffic pattern.
10. A method of making or using a computer system with a unified storage, processing and network switching fabrics, comprising the steps of:
providing a plurality of processing modules, each comprising a plurality of processing nodes, each processing node is connected to all remaining ones of the processing nodes within the module in which that processing node is comprised, to form an intra-module connection;
providing, for at least one of the processing nodes, an inter-module port;
for each of the inter-module ports, determining, for the purpose of making an inter-module connection, another one of the processing nodes in the processing module different from the processing module in which that processing node is comprised; and
establishing the inter-module connection based upon the determination by establishing the inter-module connection between the processing modules, wherein the number of the processing modules is M and each of the processing modules has at least P inter-module ports, connecting a port P on the processing modules (2 P *N+1) through (2 P *N+2 P-1 ) to a further port P on the processing modules (2 P *N+2 P-1 +1) through (2 P *N+2), respectively, where N is [0, 1, . . . , (M/2 P )−1], and P is [1, 2, . . . , P].
11. A method according to claim 10 , further comprising the step of:
forming the intra-module connection via at least of a bus, a ring, a star, a mesh, and a crossbar switch.
12. A method according to claim 10 , further comprising the step of:
providing each of the processing nodes as one of a physical node and a virtual node.
13. A method according to claim 10 , wherein each of the processing nodes comprises at least one of a processing element, a memory controller, a memory, a storage controller, a storage device, and interfaces to intra-module and inter-module ports.
14. A method according to claim 10 , further comprising the steps of:
establishing a further inter-module connection, comprising:
linearly connecting a series of the remaining processing modules; and
connecting the beginning and the end of the series to form a ring;
wherein the connections are provided using two inter-module ports from each of the connected processing modules.
15. A method according to claim 14 , further comprising the step of:
connecting one unused inter-module port on one of the processing modules on the ring to another unused inter-module port on another, non-adjacent processing module on the ring.
16. A method according to claim 14 , further comprising at least one of the steps of:
providing a connection between a processing module N on the ring to a processing module N+S on the ring, by connecting one unused inter-module port on the processing module N to another unused inter-module port on the processing module N+S, wherein M is the number of the processing modules on the ring, S is an integer divisor of M, N is [x, x+S, . . . , x+(M−S), increment by S], and x is [1, 2, . . . , S−1];
providing a connection between a processing module N on the ring to a processing module N+S−1 on the ring, by connecting one unused inter-module port on the processing module N to another unused inter-module port on the processing module N+S−1, wherein M is the number of the processing modules on the ring, S is an integer divisor of M, N is [x, x+S, . . . , x+(M−S), increment by S], and x is [1, 2, . . . , S−1]; and
providing a connection between a processing module N on the ring to a processing module N+S−r on the ring, by connecting one unused inter-module port on the processing module N to another unused inter-module port on the processing module N+S−r, wherein M is the number of the processing modules on the ring, S is an integer divisor of M, N is [x, x+S, . . . , x+(M−S), increment by S], x is [1, 2, . . . , S−1], and r is [2, 3, . . . , S−1].
17. A method according to claim 10 , further comprising the step of:
transferring a data packet from one of the processing nodes to another one of the processing nodes through the established inter-module connection.
18. A method according to claim 17 , further comprising the steps of:
measuring the data packet traffic pattern; and
amending the established inter-module connection based upon the packet traffic pattern.