Distributed file system metering and hardware resource usage

Patent metadata

Field	Value
Publication number	US-10691647-B2
Application number	US-201816148336-A
Country	US
Kind code	B2
Filing date	Oct 1, 2018
Priority date	Aug 16, 2016
Publication date	Jun 23, 2020
Grant date	Jun 23, 2020

Abstract

A method and system for automatically metering a distributed file system node is provided. The method includes receiving data associated with jobs for execution via a distributed file system. Characteristics of the jobs are uploaded and policy metrics data associated with hardware usage metering is retrieved. Resource requests associated with hardware resource usage are retrieved and attributes associated with the resource requests are uploaded. The policy metrics data is analyzed and a recommendation circuit is queried with respect to the resource requests. A set of metrics of the policy metrics data associated with the resource requests is determined and a machine learning circuit is updated. Utilized hardware resources are determined with respect to the hardware usage metering and said resource requests.

First claim

What is claimed is:

1. A distributed file system metering and hardware usage technology improvement method comprising:
receiving from a user, by a processor of a hardware device comprising specialized discrete non-generic analog, digital, and logic based plugin circuitry including a specially designed integrated circuit designed for only implementing said distributed file system metering and hardware usage technology improvement method, job data associated with jobs for execution via a distributed file system;
retrieving, by said processor enabling a policy engine circuit of said hardware device, policy and cost metrics data associated with hardware usage metering, wherein said policy and cost metrics data comprises policies implemented as pluggable components defined in advance, by: uploading xml files, via descriptor files or j son files, or via a command line, for said jobs;
retrieving, by said processor enabling a hardware device cluster, resource requests describing hardware resource usage of said jobs with respect to a level of the actual utilization of hardware resources;
querying, by said processor enabling a job descriptor engine of said hardware device, a first circuit for locating said policy and cost metrics data of said resource requests;
updating, by said processor based on a set of metrics of said policy and cost metrics data associated with said resource requests, a machine learning circuit with characteristics of said jobs associated with network delays, sudden change of memory data volumes, and hardware device failure;
detecting, by said processor enabling voltage sensors comprised by said hardware device, voltages associated with a hardware cluster;
detecting, by said processor enabling temperature sensors comprised by said hardware device, temperature readings associated with said hardware cluster, wherein said voltages and said temperature readings are analyzed to indicate a speed and usage of CPUs and a memory space available for said hardware cluster;
determining, by said processor enabling a metrics circuit with respect to said set of metrics, said voltages, and said temperature readings, utilized hardware resources of said hardware cluster with respect to said hardware usage metering of said resource requests; and
allocating, by said processor based on results of said characteristics of said jobs and results of said determining said utilized hardware resources, specified queue memory and associated processor cores to said hardware cluster thereby enabling control functionality for overall resource utilization of said hardware cluster and improving execution of said jobs being executed on said hardware cluster;
generating, by said processor executing a resource descriptor circuit, resource infrastructure associated with hardware resources of said resource requests, wherein said generating is executed based on said policy engine circuit providing various policies and cost metrics for calculating said hardware usage metering, and wherein said policy engine circuit comprises the pluggable components;
monitoring, by said processor executing said job descriptor engine, a plurality of nodes, hardware resources, processors, memory devices, and distributed system networks; and
determining, by said processor, a configuration of said hardware device cluster based on an analysis of types of workloads and said characteristics of said jobs;
determining, by said processor, workloads associated with said plurality of nodes, said hardware resources, said processors, said memory devices, and said distributed system networks;
determining, by said processor, a recommendation for improving said configuration of said hardware device cluster; and
calculating, by said computer processor, a cost for improving said configuration of said hardware device cluster,
wherein said Cost=(Confidence Factor)*Σ 0 n ((a Σ 0% 100% Memory)+(b Σ 0% 100% CPU Utilization)+(c Σ n=1 n=∞ Network Data Transfer)+(d ∫ t=0 t=n1 Scheduler Time)+(e Σ 0 n2 Preemption)+(fΣ 0 n3 Slots Used)), wherein n is a number of said plurality of nodes, n1 is an execution time of said jobs, n2 is a number of times said jobs were preempted, and n3 is a number of slots allocated to said job, wherein a, b, c, d, and e are weightage factors, wherein (a) comprises a weightage factor associated with memory, wherein said weightage factor associated with said memory comprises a variable per unit cost associated with said memory, wherein (b) comprises a weightage factor for a CPU, wherein said weightage factor for said CPU comprises a variable per unit cost associated with said CPU, wherein (c) comprises a weightage factor for network data transfer, wherein said weightage factor for said network data transfer comprises a variable per unit cost associated with said network data transfer, wherein (d) comprises a weightage factor associated with scheduling processes utilized for scheduling said jobs, wherein (e) comprises a weightage factor for a total number of preemptions, wherein said weightage factor for said total number of preemptions comprises a variable per unit cost associated with said total number of preemptions, wherein (f) comprises a total number of slots occupied and used, and wherein said (a), (b), (c), (d), (e), and (f) remain constant for a same type or category of said jobs if all additional parameters of a distributed environment remain constant.

2. The method of claim 1 , wherein said distributed file system comprises a multi tenanted distributed file system.

3. The method of claim 2 , wherein said job descriptor engine comprises circuitry for defining said characteristics of said jobs for execution on: said plurality of nodes, said hardware resources, said processors, said memory devices, and said distributed system networks.

4. The method of claim 3 , wherein said job descriptor engine receives inputs associated with said characteristics of said jobs during submission of said jobs submission via a descriptive syntax, wherein said characteristics of said jobs comprise a minimum and maximum memory required or a limit associated with a data transfer rate.

5. The method of claim 4 , wherein said job descriptor engine executes said machine learning circuit for dynamically updating itself with a history of job types, of said jobs, that have been executed in a specified sequence or have been submitted based on similar groups of users.

6. The method of claim 4 , wherein said job descriptor engine dynamically updates said characteristics of the jobs based on network delays, sudden change of data volumes, or hardware failure, and wherein said job descriptor engine monitors said plurality of nodes, said hardware resources, said processors, a cluster health, and cluster optimum processing capability.

7. The method of claim 5 , further comprising:
segmenting, by said processor, data describing said jobs for allocation of resources and estimation of capacity requirements.

8. The method of claim 7 , further comprising:
generating, by said processor, dynamic values for constants associated with the cost function.

9. The method of claim 1 , wherein said resource infrastructure enables resource infrastructure improvements comprising replacing hardware components of said hardware resources.

10. The method of claim 1 , wherein said resource infrastructure enables resource infrastructure improvements comprising adding new hardware resources to said resource infrastructure comprising said hardware resources.

11. The method of claim 1 , wherein said resource infrastructure enables resource infrastructure improvements comprising optimizing functionality resource utilization of said hardware resources with respect to each of

Assignees

• IBM

Inventors

• Badjatia Sudeep
• Bhose Rajarshi
• Deb Bijoy
• Nandi Ripon

Classifications

• H04W4/24 Primary

  Accounting or billing · CPC title

• H04L12/14

  Charging, metering or billing arrangements specially adapted for data communications, e.g. authentication, authorisation and accounting [AAA] framework · CPC title

• G06F11/3442

  for planning or managing the needed capacity · CPC title

• G06F11/3447

  Performance evaluation by modeling · CPC title

• G06F16/183 Primary

  Provision of network file services by network file servers, e.g. by using NFS, CIFS (network file access protocols H04L67/1097) · CPC title