Expandable distributed core architectures having reserved interconnect bandwidths
View Patent ↗A network system for an expandable distributed core architecture (“DCA”) includes multiple spine switches, multiple leaf switches coupled thereto, and a plurality of interconnections coupling each of the spine switches to each of the leaf switches. Servers, storage, and other system items and resources can be coupled to the leafs via downlinks and uplinks. Each of the plurality of spine switches includes at least one unused spine port that is reserved for a future expansion of the DCA in order to accommodate the addition of at least one other separate leaf switch to the at least one unused spine port. Additional ports can be unused and reserved for future expansion additions of more leafs and/or more spines. Also, a computing system assists in the design of the specific number and configuration of spines, leafs and interconnections based upon user inputs regarding current and future system needs.
1. A method of implementing an expandable distributed core architecture (“DCA”) having reserved interconnect bandwidths, the method comprising:
obtaining, by a computing system, requirements for DCA operation, the requirements comprising first requirements for the expandable DCA and second requirements for an expanded DCA, the first requirements specifying a number of uplink ports and a number of downlink ports for the expandable DCA, the second requirements specifying a number of uplink ports and a number of downlink ports for the expanded DCA, wherein at least one of (i) and (ii) is true:
(i) the number of uplink ports for the expanded DCA is higher than the number of uplink ports for the expandable DCA;
(ii) the number of downlink ports for the expanded DCA is higher than the number of downlink ports for the expandable DCA; and
determining, by the computing system, a configuration of the expandable DCA based on the first and second requirements and on an optimality objective, wherein determining the configuration comprises using the numbers of the uplink and downlink ports for the expandable DCA and the numbers of the uplink and downlink ports for the expanded DCA to determine a number of spine switches for the expandable DCA, a number of leaf switches for the expandable DCA, and direct interconnections between the spine and leaf switches for the expandable DCA;
wherein the configuration is for the expandable DCA corresponding to the first requirements but not the second requirements, but the configuration takes into account the second requirements to provide expandability of the expandable DCA to the expanded DCA with the expanded DCA including all of the spine and leaf switches of the expandable DCA and all of the direct interconnections between the spine and leaf switches of the expandable DCA, the configuration requiring at least one of the spine and leaf switches of the expandable DCA to have at least one port that is unused in the expandable DCA but is used in the expanded DCA;
wherein the optimality objective comprises at least one of:
low cost;
high throughput;
high speed;
minimal amount of switches;
minimal amount of routing components.
2. The method of claim 1 , wherein the expandable DCA configuration includes at least one direct interconnection between every spine switch and every leaf switch.
3. The method of claim 1 wherein each of the expandable and expanded DCAs includes at least one direct interconnection between every spine switch and every leaf switch.
4. The method of claim 1 wherein each spine and leaf switch of the expandable DCA has a port that is unused in the expandable DCA but is used in the expanded DCA.
5. The method of claim 1 wherein determining said configuration comprises determining a configuration that reduces a cost of each of the expandable and expanded DCAs.
6. The method of claim 1 wherein determining said configuration comprises determining a configuration that increases throughput of each of the expandable and expanded DCAs.
7. The method of claim 1 wherein determining said configuration comprises determining a configuration that increases speed of each of the expandable and expanded DCAs.
8. The method of claim 1 wherein determining said configuration comprises determining a configuration that reduces a number of the spine and leaf switches in each of the expandable and expanded DCAs.
9. The method of claim 1 wherein determining said configuration comprises determining a configuration that reduces a number of routing components in each of the expandable and expanded DCAs.
10. The method of claim 1 wherein each of the first and second requirements comprises a bandwidth requirement.
11. The method of claim 10 wherein the bandwidth requirement is higher for the expanded DCA than for the expandable DCA.
12. The method of claim 1 wherein each of the first and second requirements comprises oversubscription ratio.
13. The method of claim 1 wherein each of the first and second requirements specifies a switch model to serve as the spine and leaf switches.
14. The method of claim 1 wherein the expanded DCA has more of the leaf switches than the expandable DCA but the same number of the spine switches as in the expandable DCA.
15. The method of claim 1 wherein the expanded DCA has more of the leaf switches and more of the spine switches than the expandable DCA.
16. A computing system comprising:
one or more processors;
one or more storage components;
one or more peripherals; and
a software portion stored in the one or more storage components which, when executed by the one or more processors, causes the computing system to perform a method for implementing an expandable distributed core architecture (“DCA”) having reserved interconnect bandwidths, the method comprising:
obtaining, by the one or more processors, requirements for DCA operation, the requirements comprising first requirements for the expandable DCA and second requirements for an expanded DCA, the first requirements specifying a number of uplink ports and a number of downlink ports for the expandable DCA, the second requirements specifying a number of uplink ports and a number of downlink ports for the expanded DCA, wherein at least one of (i) and (ii) is true:
(i) the number of uplink ports for the expanded DCA is higher than the number of uplink ports for the expandable DCA;
(ii) the number of downlink ports for the expanded DCA is higher than the number of downlink ports for the expandable DCA; and
determining, by the one or more processors, a configuration of the expandable DCA based on the first and second requirements and on an optimality objective, wherein determining the configuration comprises using the numbers of the uplink and downlink ports for the expandable DCA and the numbers of the uplink and downlink ports for the expanded DCA to determine a number of spine switches for the expandable DCA, a number of leaf switches for the expandable DCA, and direct interconnections between the spine and leaf switches for the expandable DCA;
wherein the configuration is for the expandable DCA corresponding to the first requirements but not the second requirements, but the configuration takes into account the second requirements to provide expandability of the expandable DCA to the expanded DCA with the expanded DCA including all of the spine and leaf switches of the expandable DCA and all of the direct interconnections between the spine and leaf switches of the expandable DCA, the configuration requiring at least one of the spine and leaf switches of the expandable DCA to have at least one port that is unused in the expandable DCA but is used in the expanded DCA;
wherein the optimality objective comprises at least one of:
low cost;
high throughput;
high speed;
minimal amount of switches;
minimal amount of routing components.
17. The computing system of claim 16 wherein each of the first and second requirements comprises a bandwidth requirement.