Migration of guest operating system optimization tool settings in a multi-hypervisor data center environment
Presented herein are embodiments for automating analytics-based migration of virtual machine optimization tool settings to different hypervisor environments. Currently, no marketplace workload migration utilities have previously dealt with this scenario. In one or more embodiments, a system gathers data and uses analytics on the data to devise one or more translation/correlation rules or models for virtual machine migration. In one or more embodiments, using historical manually settings, correlations can be determined. Given data about a source guest operating system (OS) tool settings, one or more translation/correlation models may be used to facilitate the translation of guest OS optimization tool settings from the source environment to the destination environment so that migrated virtual machine functions the same as or nearly the same as it did on the source hypervisor.
1. A computer-implemented method for determining settings for migration of a virtual machine comprising:
given a virtual machine of a first type comprising a guest operating system (OS) operating on a source hypervisor that has been selected for migration to a virtual machine of a second type operating on a destination hypervisor that is of a different type than the source hypervisor:
gathering data regarding source guest OS tool settings, which help improve performance of the source guest OS in the source hypervisor's virtualized environment by affecting at least one of data exchange and time synchronization, about the virtual machine operating on the source hypervisor;
obtaining, from a repository, a set of settings correlation models for a direction-specific migration from the source hypervisor to the destination hypervisor;
using at least some of the gathered data and at least one or more correlation models from the set of setting correlation models for the direction-specific migration, determining a set of correlated settings for the direction-specific migration of the virtual machine from the source hypervisor to the destination hypervisor; and
applying a finalized set of correlated settings obtained from the set of correlated settings to a destination guest OS tool for the virtual machine.
2. The computer-implemented method of claim 1 wherein the finalized set of correlated settings are obtained from the set of correlated settings by performing at least one step comprising:
using the set of correlated settings as the finalized set of correlated settings;
updating the set of correlated settings using one or more pre-defined rules that override at least one of the correlated settings in the set of correlated settings;
updating the set of correlated settings using one or more user-supplied inputs; and
updating one or more values for one or more correlated settings in the set of correlated settings.
3. The computer-implemented method of claim 1 further comprising:
gathering data regarding source hypervisor-related settings, data regarding destination hypervisor-related settings, or both; and
using at least some the gathered data in determining one or more implementation-related settings for the destination hypervisor for the direction-specific migration.
4. The computer-implemented method of claim 1 further comprising:
for each virtual machine from a set of migrated virtual machines:
recording source guest OS tool settings data about the virtual machine operating on the source hypervisor; and
recording destination guest OS tool settings data about the virtual machine operating on the destination hypervisor after migration;
using the source guest OS tool settings data and the corresponding destination guest OS tool settings data for the set of migrated virtual machines as training data to develop at least some of the set of settings correlation models for the direction-specific migration using one or more machine learning techniques; and
storing the at least some of the set of settings correlation models for the direction-specific migration in a datastore.
5. The computer-implemented method of claim 4 further comprising:
recording monitored settings-related information related to the source hypervisor, the destination hypervisor, or both; and
using at least some of the monitored settings-related information when developing at least some of the set of settings correlation models for the direction-specific migration.
6. The computer-implemented method of claim 4 wherein one of the one or more machine learning techniques is a Pearson correlation coefficient that correlates one or more settings in the source guest OS tool settings to one or more settings in the destination guest OS tool settings.
7. The computer-implemented method of claim 1 wherein the step of obtaining, from a repository, a set of settings correlation models for the direction-specific migration from the source hypervisor to the destination hypervisor comprising:
determining whether to obtain the set of settings correlation models from a local datastore or a centralized datastore;
responsive to determining to obtain the set of settings correlation models from the local datastore, accessing the set of settings correlations models from the local datastore; and
responsive to determining to obtain the set of settings correlation models from the centralized datastore, accessing the set of settings correlations models from the centralized datastore.
8. A computer-implemented method for determining settings for migration of a virtual machine comprising:
for each virtual machine from a set of migrated virtual machines:
recording source guest operating system (OS) tool settings data about the virtual machine operating on a source hypervisor, in which the source guest OS tool settings help improve performance of the source guest OS in a source hypervisor environment by affecting at least one of data exchange and time synchronization; and
recording destination guest OS tool settings data about the virtual machine operating on a destination hypervisor that is of a different type than the source hypervisor after migration of the virtual machine to the destination hypervisor, in which the destination guest OS tool settings help improve performance of the destination guest OS in a destination hypervisor environment by affecting at least one of data exchange and time synchronization;
using the source guest OS tool settings data and the corresponding destination guest OS tool settings data for the set of migrated virtual machines as training data to develop a set of settings correlation models for a direction-specific migration using one or more machine learning techniques, in which the set of setting correlation models correlate one or more settings of the source guest OS tool for a virtual machine operating on a first hypervisor to one or more settings of the destination guest OS tool for the virtual machine operating on a second hypervisor after migration; and
storing the at least some of the set of settings correlation models for the direction-specific migration in a datastore.
9. The computer-implemented method of claim 8 further comprising:
recording monitored settings-related information related to the source hypervisor, the destination hypervisor, or both; and
using at least some of the monitored settings-related information when developing at least some of the set of settings correlation models for the direction-specific migration.
10. The computer-implemented method of claim 8 wherein one of the one or more machine learning techniques is a Pearson correlation coefficient that correlates one or more settings in the source guest OS tool settings to one or more settings in the destination guest OS tool settings.
11. The computer-implemented method of claim 10 further comprising:
responsive to a Pearson correlation coefficient not achieving an acceptable threshold value:
obtaining additional data of migrated virtual machines; and
correlating one or more settings of the source guest OS tool settings to one or more settings of the destination guest OS tool settings.
12. The computer-implemented method of claim 8 wherein one of the one or more machine learning techniques comprises using a neural network and the method further comprises:
for each virtual machine from a set of migrated virtual machines, using at least some of the source guest OS tool settings data for a migrated virtual machine as an input feature and using the corresponding destination guest OS tool settings data as ground truth data;
repeating until a stop condition is reached:
inputting a set of input features into a neural network model to predict correlated settings for one or more settings in the destination guest OS tool;
using a comparison of the predicted correlates settings relative to the corresponding destination guest OS tool settings data for the migrated virtual machines to determine an update for the neural network; and
updating one or more parameters of the neural network using the update; and
outputting the neural network for predicting destination guest OS tool settings given source guest OS tool settings data.
13. The computer-implemented method of claim 12 further comprises:
responsive to the neural network not obtaining sufficient accuracy in predicting:
obtaining additional training data of migrated virtual machines; and
repeating the steps comprising:
for each virtual machine from the set of migrated virtual machines, using at least some of the source guest OS tool settings data for a migrated virtual machine as an input feature and using the corresponding destination guest OS tool settings data as ground truth data;
repeating until a stop condition is reached:
inputting a set of input features into a neural network model to predict correlated settings for one or more settings in the destination guest OS tool;
using a comparison of the predicted correlates settings relative to the corresponding destination guest OS tool settings data for the migrated virtual machines to determine an update for the neural network; and
updating one or more parameters of the neural network using the update; and
outputting the neural network for predicting destination guest OS tool settings given source guest OS tool settings data.
14. The computer-implemented method of claim 8 wherein the recorded source guest operating system (OS) tool settings data and the recorded destination guest OS tool settings data is obtained from a plurality of organizations.
15. The computer-implemented method of claim 8 further comprising:
given a virtual machine comprising a guest operating system (OS) operating on a source hypervisor that has been selected for migration to a destination hypervisor:
gathering data regarding source guest OS tool settings about the virtual machine operating on the source hypervisor;
obtaining, from a repository, at least some of the set of settings correlation models for the direction-specific migration from the source hypervisor to the destination hypervisor;
using at least some of the gathered data and at least one or more correlation models from the set of setting correlation models for the direction-specific migration, determining a set of correlated settings for the direction-specific migration of the virtual machine from the source hypervisor to the destination hypervisor; and
applying a finalized set of correlated settings obtained from the set of correlated settings to a destination guest OS tool for the virtual machine.
16. A system comprising:
one or more processors; and
a non-transitory computer-readable medium or media comprising one or more sets of instructions which, when executed by at least one of the one or more processors, causes steps to be performed comprising:
given a virtual machine of a first type comprising a guest operating system (OS) operating on a source hypervisor that has been selected for migration to a virtual machine of a second type operating on a destination hypervisor that is of a different type than the source hypervisor:
gathering data regarding source guest OS tool settings about the virtual machine operating on the source hypervisor, in which the source guest OS tool settings help improve performance of the source guest OS in a source hypervisor environment by affecting at least one of data exchange and time synchronization;
obtaining, from a repository, at least some of a set of settings correlation models for a direction-specific migration from the source hypervisor to the destination hypervisor;
using at least some of the gathered data and at least one or more correlation models from the set of setting correlation models for a direction-specific migration, determining a set of correlated settings for the direction-specific migration of the virtual machine from the source hypervisor to the destination hypervisor; and
applying a finalized set of correlated settings obtained from the set of correlated settings to a destination guest OS tool for the virtual machine.
17. The system of claim 16 wherein the finalized set of correlated settings are obtained from the set of correlated settings by performing at least one step comprising:
using the set of correlated settings as the finalized set of correlated settings;
updating the set of correlated settings using one or more pre-defined rules that override at least one of the correlated settings in the set of correlated settings;
updating the set of correlated settings using one or more user-supplied inputs; and
updating one or more values for one or more correlated settings in the set of correlated settings.
18. The system of claim 16 further comprising:
gathering data regarding source hypervisor-related settings, data regarding the destination hypervisor-related settings, or both; and
using at least some the gathered data in determining one or more implementation-related settings for the destination hypervisor for the direction-specific migration.
19. The system of claim 16 further comprising:
for each virtual machine from a set of migrated virtual machines:
recording source guest OS tool settings data about the virtual machine operating on a source hypervisor; and
recording destination guest OS tool settings data about the virtual machine operating on a destination hypervisor after migration;
using the source guest OS tool settings data and the corresponding destination guest OS tool settings data for the set of migrated virtual machines as training data to develop at least some of the set of settings correlation models for the direction-specific migration using one or more machine learning techniques; and
storing the at least some of the set of settings correlation models for direction-specific migration in a datastore.
20. The system of claim 19 wherein one of the one or more machine learning techniques is a Pearson correlation coefficient that correlates one or more settings in the source guest OS tool settings to one or more settings in the destination guest OS tool settings.