Adaptive multi-tenant monitoring in resource constrained environments

Patent metadata

Field	Value
Publication number	US-10555142-B2
Application number	US-201715700043-A
Country	US
Kind code	B2
Filing date	Sep 8, 2017
Priority date	Sep 8, 2017
Publication date	Feb 4, 2020
Grant date	Feb 4, 2020

Abstract

An example method for multi-tenant adaptive monitoring comprises detecting occurrence of a trigger event and modifying a selection of metrics included in a plurality of monitored metrics that are monitored using available resources of a plurality of tenants. The method further comprises assigning a respective monitoring frequency for each metric; computing respective weights for each metric in the modified selection of metrics; performing a feasibility check to find a solution to a mathematical model for monitoring the modified selection of metrics at the respective assigned monitoring frequency for each metric; and, in response to determining that a solution to the mathematical model cannot be found, adjusting the respective monitoring frequency for one or more metrics. The method further comprises, in response to finding a first solution to the mathematical model, allocating processing associated with monitoring each metric among the available resources of the plurality of tenants.

First claim

What is claimed is:

1. A computer-implemented method of multi-tenant adaptive monitoring, the method comprising:
detecting occurrence of a trigger event;
modifying, in response to detecting occurrence of the trigger event, a selection of metrics included in a plurality of monitored metrics that are monitored using available resources of a plurality of tenants, each metric associated with at least one of the plurality of tenants, wherein the modified selection of metrics includes at least one shared metric associated with at least two or more of the plurality of tenants;
assigning a respective monitoring frequency for each metric included in the modified selection of metrics, wherein assigning the respective monitoring frequency for the at least one shared metric includes computing the respective monitoring frequency as a function of a respective desired monitoring frequency for each of the two more tenants associated with the at least one shared metric;
computing respective weights for each metric in the modified selection of metrics;
performing a feasibility check to find a solution to a mathematical model for monitoring the modified selection of metrics at the respective assigned monitoring frequency for each metric using the available resources of the plurality of tenants;
in response to determining that a solution to the mathematical model cannot be found, adjusting the respective monitoring frequency for one or more metrics in the modified selection of metrics based, at least in part, on the respective weight for each metric and on the at least one tenant associated with each metric until a solution to the mathematical model is found; and
in response to finding a first solution to the mathematical model, allocating processing associated with monitoring each metric in the modified selection of metrics among the available resources of the plurality of tenants based on the respective monitoring frequency of each metric and based on the at least one tenant associated with each metric.

2. The computer-implemented method of claim 1 , wherein the detected trigger event is one of an environment-based event or a metric-based event.

3. The computer-implemented method of claim 1 , wherein the detected trigger event is a tenant-preference-based event; and
wherein detecting occurrence of the tenant-preference-based event includes predicting a change in a tenant preference based on an analysis of historical data.

4. The computer-implemented method of claim 1 , wherein allocating processing associated with monitoring each metric in the modified selection of metrics further comprises proportionally distributing processing associated with monitoring each respective metric in the modified selection of metrics amongst the available resources of the plurality of tenants according to a respective minimum frequency for monitoring the respective metric desired by each respective tenant requesting the respective metric.

5. The computer-implemented method of claim 1 , wherein adjusting the respective monitoring frequency for one or more metrics in the modified selection of metrics includes:
initializing a first set of metrics, a second set of metrics, and a set of unused resources; wherein the first set of metrics and the second set of metrics are initialized as empty and the set of unused resources is initialized to include all the available resources of the plurality of tenants;
ranking the metrics in the modified selection of metrics in a descending order based on the respective weight of each metric in the modified selection of metrics;
iteratively performing a respective feasibility check for each respective metric in the modified selection of metrics beginning with a highest ranked metric, wherein each respective feasibility check is based on each respective metric and on metrics in the first set of metrics;
for each respective metric for which the respective feasibility check indicates a solution exists for monitoring the respective metric and the metrics in the first set of metrics at the respective computed monitoring frequency for each metric using the available resources of the plurality of tenants, updating the first set of metrics to include the respective metric and updating the set of unused resources to remove resources needed for monitoring the respective metric;
for each respective metric for which the respective feasibility check cannot find a solution, updating the second set of metrics to include the respective metric; and
for each respective tenant associated with at least one metric in the second set of metrics, reducing the respective monitoring frequency of all metrics associated with the respective tenant.

6. The computer-implemented method of claim 1 , wherein allocating processing associated with monitoring each metric in the modified selection of metrics includes:
ranking the metrics in the modified selection of metrics in a descending order based on the respective weight of each metric in the modified selection of metrics;
incrementally increasing the respective monitoring frequency of each metric in the modified selection of metrics to utilize unused resources of the plurality of tenants;
wherein the respective monitoring frequency of each metric in the modified selection of metrics is incrementally increased one metric at a time, beginning with the highest ranked metric, until a respective feasibility check performed after the incremental increase of the respective monitoring frequency of each metric indicates that a solution cannot be found for monitoring the metrics in the modified selection of metrics at the respective increased monitoring frequency for the respective metric using the available resources of the plurality of tenants.

7. The computer-implemented method of claim 1 , wherein the modified selection of metrics includes at least one unshared metric that is associated with a single tenant of the plurality of tenants, wherein the processing associated with monitoring the at least one unshared metric is allocated only to the respective available resources of the single tenant associated with the unshared metric.

8. A computer program product comprising a computer readable storage medium having a computer readable program stored therein, wherein the computer readable program, when executed by a processor, causes the processor to:
detect occurrence of a trigger event;
modify, in response to detecting occurrence of the trigger event, a selection of metrics included in a plurality of monitored metrics that are monitored using available resources of a plurality of tenants, each metric associated with at least one of the plurality of tenants, wherein the modified selection of metrics includes at least one shared metric associated with at least two or more of the plurality of tenants;
assign a respective monitoring frequency for each metric included in the modified selection of metrics, wherein assigning the respective monitoring frequency for the at least one shared metric includes computing the respective monitoring frequency as a function of a respective desired monitoring frequency for each of the two more tenants associated with the at least one shared metric;
compute respective weights for each metric in the modified selection of metrics;
perform a feasibility check to find a solution to a mathematical model for monitoring the modified selection of metrics at the respective computed monitoring frequency for each metric using the available resources of the plurality of tenants;
in response to determining that a solution to the mathematical model cannot be found, adjust the respective monitoring frequency for one or more metrics in the modified selection of metrics based, at least in part, on the respective weight for each m

Assignees

• IBM

Inventors

• Tata Samir
• Mohamed Mohamed
• Megahed Aly

Classifications

• H04L43/08

  Monitoring or testing based on specific metrics, e.g. QoS, energy consumption or environmental parameters · CPC title

• H04L67/12

  specially adapted for proprietary or special-purpose networking environments, e.g. medical networks, sensor networks, networks in vehicles or remote metering networks · CPC title

• H04W4/38 Primary

  for collecting sensor information · CPC title

• G06N7/01

  Probabilistic graphical models, e.g. probabilistic networks · CPC title

• G06N7/005

  Physics · mapped topic