Technologies for performance inspection at an endpoint node

Patent metadata

Field	Value
Publication number	US-10135708-B2
Application number	US-201514866536-A
Country	US
Kind code	B2
Filing date	Sep 25, 2015
Priority date	Sep 25, 2015
Publication date	Nov 20, 2018
Grant date	Nov 20, 2018

Abstract

Technologies for monitoring communication performance of a high performance computing (HPC) network include a performance probing engine of a source endpoint node of the HPC network. The performance probing engine is configured to generate a probe request that includes a timestamp of the probe request and transmit the probe request to a destination endpoint node of the HPC network communicatively coupled to the source endpoint node via the HPC network. The performance probing engine is additionally configured to receive a probe response from the destination endpoint node via the HPC network and to generate another timestamp that corresponds to the probe request having been received. Further, the performance probing engine is configured to determine a round-trip latency as a function of the probe request and probe response timestamps. Other embodiments are described and claimed.

First claim

The invention claimed is:

1. A source endpoint node of a network to monitor communication performance of the network, the source endpoint node comprising:
one or more processors; and
one or more memory or data storage devices having stored therein a plurality of instructions that, when executed by the one or more processors, cause the source endpoint node to:
generate a probe request and an identifier of the probe request, wherein the probe request includes performance metadata of the source endpoint node, and wherein the performance metadata indicates at least one of power consumption, bandwidth, throughput, or latency of the source endpoint node;
generate, using a clock of the source endpoint node, a first timestamp;
store the identifier of the probe request and the first timestamp as a key-value pair in a local storage of the source endpoint node; transmit the probe request to a destination endpoint node of the network communicatively coupled to the source endpoint node via the network;
receive, in response to having transmitted the probe request, a probe response from the destination endpoint node via the network;
generate, using the clock and in response to having received the probe request, a second timestamp;
determine a round-trip latency as a function of the first and second timestamps; determine that the round-trip latency is indicative of a degradation in a performance level of the network;
increment, in response to a determination that the identifier is no longer in the local storage, an over-time counter;
determine one or more points of congestion in the network based on a plurality of historical values associated with the over-time counter; and
perform a remedial action in response to the determination that the round-trip latency is indicative of the degradation in the performance level of the network.

2. The source endpoint node of claim 1 ,
wherein to transmit the probe request includes to transmit the identifier with the probe request, wherein to receive the probe response includes to receive the identifier with the probe response, and wherein to determine the round-trip latency as a function of the first and second timestamps includes to retrieve the first timestamp associated with the probe request based on the identifier of the probe request.

3. The source endpoint node of claim 2 , wherein the plurality of instructions, when executed by the one or more processors, further cause the source endpoint node to:
discard the probe response.

4. The source endpoint node of claim 1 , wherein the plurality of instructions, when executed by the one or more processors, further cause the source endpoint node to include the first timestamp into the probe request, wherein to transmit the probe request to the destination endpoint node of the network comprises to transmit the probe request with the first timestamp, and wherein to receive the probe response from the destination endpoint node comprises to receive the probe response from the destination endpoint node that includes the first timestamp.

5. The source endpoint node of claim 1 , wherein the plurality of instructions, when executed by the one or more processors, further cause the source endpoint node to include a request for remote performance metadata of the destination endpoint node into the probe request, and wherein to receive the probe response from the destination endpoint node includes to receive remote performance metadata of the destination endpoint node.

6. The source endpoint node of claim 5 , wherein the plurality of instructions, when executed by the one or more processors, further cause the source endpoint node to determine one or more performance characteristics of the network as a function of the remote performance metadata.

7. The source endpoint node of claim 6 , wherein the remedial action is performed in response to a further determination that a performance characteristic of the network is indicative of the degradation in the performance level of the network.

8. The source endpoint node of claim 1 , wherein the plurality of instructions, when executed by the one or more processors, further cause the source endpoint node to:
retrieve the performance metadata of the source endpoint node, wherein the performance metadata of the source endpoint node is usable by the destination endpoint node to determine a performance characteristic of the network.

9. The source endpoint node of claim 1 , wherein to transmit the probe request to the destination endpoint node comprises to (i) include the probe request into an outgoing data packet and (ii) transmit the outgoing data packet including the probe request.

10. One or more non-transitory computer-readable storage media comprising a plurality of instructions stored thereon that in response to being executed cause a source endpoint node to:
generate a probe request and an identifier of the probe request, wherein the probe request includes performance metadata of the source endpoint node, and wherein the performance metadata indicates at least one of power consumption, bandwidth, throughput, or latency of the source endpoint node;
generate, using a clock of the source endpoint node, a first timestamp;
store the identifier of the probe request and the first timestamp as a key-value pair in a local storage of the source endpoint node;
transmit the probe request to a destination endpoint node of the network communicatively coupled to the source endpoint node via the network;
receive, in response to having transmitted the probe request, a probe response from the destination endpoint node via the network;
generate, using the clock and in response to having received the probe request, a second timestamp;
determine a round-trip latency as a function of the first and second timestamps;
determine that the round-trip latency is indicative of a degradation in a performance level of the network;
increment, in response to a determination that the identifier is no longer in the local storage, an over-time counter;
determine one or more points of congestion in the network based on a plurality of historical values associated with the over-time counter; and
perform a remedial action in response to the determination that the round-trip latency is indicative of the degradation in the performance level of the network.

11. The one or more non-transitory computer-readable storage media of claim 10 , wherein to transmit the probe request includes to transmit the identifier with the probe request, wherein receiving the probe response includes receiving the identifier with the probe response, and wherein to determine the round-trip latency as a function of the first and second timestamps includes retrieving the first timestamp associated with the probe request based on the identifier of the probe request.

12. The one or more non-transitory computer-readable storage media of claim 11 , further comprising a plurality of instructions that in response to being executed cause the source endpoint node to: discard the probe response.

13. The one or more non-transitory computer-readable storage media of claim 10 , further comprising a plurality of instructions that in response to being executed cause the source endpoint node to include the first timestamp into the probe request, wherein to transmit the probe request to the destination endpoint node of the network comprises to transmit the probe request with the first timestamp, and wherein to receive the probe response from the destination endpoint node comprises to receive the probe response from the destination endpoint node that includes the first timestamp.

14. The one or mor

Assignees

• Intel Corp

Inventors

• Dinan James
• Keppel David

Classifications

• H04L43/10

  Active monitoring, e.g. heartbeat, ping or trace-route · CPC title

• H04L43/0876 Primary

  Network utilisation, e.g. volume of load or congestion level · CPC title

• H04L67/10 Primary

  in which an application is distributed across nodes in the network (software deployment G06F8/60; multiprogramming arrangements G06F9/46) · CPC title

• H04L43/12

  Network monitoring probes · CPC title

• H04L69/40

  for recovering from a failure of a protocol instance or entity, e.g. service redundancy protocols, protocol state redundancy or protocol service redirection (management of faults, events, alarms or notifications in data switching networks H04L41/06) · CPC title