Migrating workloads across host computing systems based on cache content usage characteristics

Patent metadata

Field	Value
Publication number	US-10255188-B2
Application number	US-201414262848-A
Country	US
Kind code	B2
Filing date	Apr 28, 2014
Priority date	Mar 17, 2014
Publication date	Apr 9, 2019
Grant date	Apr 9, 2019

Abstract

Techniques for migrating workloads across host computing systems in a virtual computing environment are described. In one embodiment, workloads executing on different host computing systems that access identical contents that are stored on storage devices are identified, with the identical contents often being cached in a cache of each of the different host computing systems. Further, migration of one or more of the identified workloads is recommended to consolidate the identified workloads on a single host computing system, thereby allowing the identical contents to be cached within the single host computing system and allowing the identified workloads to access the cached identical contents from the single host computing system after migration in accordance with the recommendation.

First claim

The invention claimed is:

1. A method of creating a workload migration recommendation comprising:
loading a global digest table (GDT) onto memory of host computing systems in a cluster;
dynamically updating hash signatures of metadata associated with contents read by workloads running on the host computing systems along with associated host computing system identifiers (IDs) in the GDT for a predetermined time interval;
identifying workloads executing on a first host computing system and a second host computing, system that access identical contents that are stored on storage devices, wherein the identical contents are often being cached in a cache of the first host computing system and the second host computing system;
determining cache content usage characteristics of the first host computing system and the second host computing system by analyzing the hash signatures of the metadata associated with the read contents along with the associated host computing system IDs in the GDT, wherein the cache content usage characteristics comprise a frequency of read requests made by the workloads for substantially same contents; and
recommending migration of at least one of the identified workloads from the first host computing system to the second host computing system based on the cache content usage characteristics, thereby allowing the identical contents to be cached within the cache of the second host computing system and allowing the migrated workloads to access the identical contents cached in the cache of the second host computing system after migration in accordance with the recommendation.

2. The method of claim 1 , wherein recommending migration of at least one of the identified workloads to consolidate the identified workloads on a single host computing system, comprises:
creating a recommendation for migrating the workloads across the host computing systems in the cluster based on the outcome of the analysis.

3. The method of claim 2 , further comprising:
repeating the steps of updating, analyzing and creating the recommendation for a next predetermined time interval.

4. The, method of claim 1 , wherein dynamically updating the hash signatures of metadata associated with the contents read by the workloads along with the associated host computing system IDs in the GDT for the predetermined time interval, comprises:
determining whether each requested content associated with a workload is in a cache of an associated host computing system using the GDT residing in memory of the associated host computing system;
if so, reading the requested contents from the cache of the associated host computing system;
if not, reading any of remaining requested contents from the storage devices, storing the read contents in the cache associated with the host computing system, and adding the hash signatures of the metadata associated with the read contents along with the associated host computing system IDs in the GDT; and
repeating the step of determining for the predetermined time interval.

5. The method of claim 4 , further comprising:
receiving read requests for the contents stored in virtual memory disk (VMDK) files located in the storage devices from the workloads associated with the host computing systems in the cluster.

6. The method of claim 1 , further comprising:
substantially simultaneously updating each of remaining CDTs upon adding a hash signature of metadata associated with current read content along with an associated host computing system ID in one of the GDTs via a cache digest trunk (CDT).

7. The method of claim 1 , further comprising:
substantially simultaneously updating each GDT by removing corresponding hash signatures and associated host computing system IDs via a cache digest trunk (CDT) upon removing any one of the host computing systems in the cluster.

8. The method of claim 1 , further comprising:
substantially simultaneously creating a replica of the GDT in a host computing system via a cache digest trunk (CDT) upon adding the host computing system in the cluster.

9. The method of claim 1 , wherein the host computing system IDs are assigned unique host computing system IDs for tagging the host computing systems in the GDT.

10. The method of claim 1 , wherein the cache content usage characteristics are locations of the workloads in the host computing systems, contents requested by the workloads and frequency of read requests made for substantially same contents.

11. The method of claim 1 , wherein the content comprises a length of 4 kilo bytes.

12. A system, comprising:
multiple host computing systems, wherein each host computing system executing multiple workloads;
storage devices communicatively coupled to the multiple host computing systems; and
a management server communicatively coupled to the multiple host computing systems, wherein the management server comprises virtual management software (VMS) to create a workload migration recommendation in a virtual computing environment, by:
loading a global digest table (GUI) onto memory of host computing systems in a cluster;
dynamically update hash signatures of metadata associated with contents read by workloads running on the host computing systems along with associated host computing system identifiers (IDs) in the GDT for a predetermined time interval;
identify workloads executing on a first host computing system and a second host computing system that access identical contents that are stored on the storage devices, wherein the identical contents are often being cached in a cache of the first host computing system and the second host computing system;
determine cache content usage characteristics of the first host computing system and the second host computing system by analyzing the hash signatures of the metadata associated with the read contents along with the associated host computing system IDs in the GDT, wherein the cache content usage characteristics comprise a frequency of read requests made by the workloads for substantially same contents; and
recommend migration of at least one of the identified workloads from the first host computing system to the second host system based on the cache content usage characteristics, thereby allowing the identical contents to be cached within the cache of the second host computing system and allowing the migrated workloads to access the identical contents cached in the cache of the second host computing system after migration in accordance with the recommendation.

13. The system of claim 12 , wherein the VMS is configured to:
create a recommendation for migrating the workloads across the multiple host computing systems in the cluster based on the outcome of the analysis.

14. The system of claim 13 , wherein the VMS is further configured to:
repeat the steps of updating, analyzing and creating the recommendation for a next predetermined time interval.

15. The system of claim 12 , wherein the VMS is configured to:
determine whether each requested content associated with a workload is in a cache of an associated host computing system using the GDT residing in the memory of the associated host computing system;
if so, read the requested contents from the cache of the associated host computing system;
if not, read any of remaining requested contents from the storage devices, storing the read contents in the cache associated with the host computing system, and adding the hash signatures of the metadata associated with the read contents along with the associated host computing system IDs in the GDT, and
repeat the step of determining for the predetermined time interval.

Assignees

• Vmware Inc

Inventors

• Manoharan Manesh Varissery
• Govindankutty Sivaprasad Kundoor
• Jose Jubish Kulathumkal

Classifications

• G06F2209/5022

  Workload threshold · CPC title

• G06F2212/452

  Instruction code · CPC title

• G06F12/0875 Primary

  with dedicated cache, e.g. instruction or stack · CPC title

• G06F9/5033 Primary

  considering data affinity · CPC title