Dynamically selecting optimal instance type for disaster recovery in the cloud
The selection of an optimal restore instance type based on a customer's speed/cost tradeoff resolution is disclosed. An automated restore activity may be performed on a baseline test VM of a predefined size using different restore instance types. The number of calibration runs or evaluations needed to identify an optimal restore instance type in terms of performance and price, with respect to bandwidth or other constraining factor, is performed on less than all of the restore instance types.
1. A method, comprising:
identifying restore instance types associated with a cloud provider, each restore instance being associated with a bandwidth tier;
performing a search on less than all of the restore instance types for each bandwidth tier, wherein the search is a binary search;
for restore instance types not included in the search and included in a calibration table, interpolating results in the calibration table for restore instance types not included in the search;
determining an optimal restore instance type from the restore instance types based on the search and based, in part, on a calibration run performed using a calibration data protection operation of a test VM; and
performing a data protection operation with a restore virtual machine based on the optimal restore instance type.
2. The method as recited in claim 1 , further comprising identifying the restore instance types and sorting the restore instance types by price.
3. The method as recited in claim 1 , further comprising performing the calibration run for each restore instance while performing the binary search.
4. The method as recited in claim 1 , further comprising applying a policy based at least on recovery price and recovery time to select a restore instance type for the data protection operation that is different from the optimal restore instance type.
5. The method of claim 4 , further comprising generating a rank table that includes a ranking for each of the restore instance types.
6. The method of claim 1 , further comprising:
creating the test VM with a VM restore instance type that matches a VM restore instance type identified during performance of the search;
using the test VM as a basis to create a test restore VM at a cloud storage site;
restoring the test VM at the cloud storage site using the test restore VM;
generating a 4-D baseline vector based on the restoration of the test VM, the 4-D baseline vector identifying a particular VM restore instance type;
generating a 5-D vector based on the 4-D baseline vector;
ranking the 5-D vector relative to one or more other 5-D vectors, all of the 5-D vectors identifying the same production site VM; and
restoring, at the cloud storage site, the production site VM identified in the 5-D vectors, and the production site VM restored at the cloud storage site has a VM restore instance type identified in the highest ranked 5-D vector.
7. A non-transitory storage medium having stored therein computer-executable instructions which, when executed by one or more hardware processors, perform the operations:
identifying restore instance types associated with a cloud provider, each restore instance being associated with a bandwidth tier;
performing a search on less than all of the restore instance types for each bandwidth tier, wherein the search is a binary search;
for restore instance types not included in the search and included in a calibration table, interpolating results in the calibration table for restore instance types not included in the search;
determining an optimal restore instance type from the restore instance types based on the search and based, in part, on a calibration run performed using a calibration data protection operation of a test VM; and
performing a data protection operation with a restore virtual machine based on the optimal restore instance type.
8. The non-transitory storage medium as recited in claim 7 , the operations further comprising identifying the restore instance types and sorting the restore instance types by price.
9. The non-transitory storage medium as recited in claim 7 , the operations further comprising performing the calibration run for each restore instance while performing the binary search.
10. The non-transitory storage medium as recited in claim 7 , the operations further comprising applying a policy based at least on recovery price and recovery time to select a restore instance type for the data protection operation that is different from the optimal restore instance type.
11. The non-transitory storage medium as recited in claim 10 , the operations further comprising generating a rank table that includes a ranking for each of the restore instance types.
12. The non-transitory storage medium as recited in claim 7 , the operations further comprising:
creating the test VM with a VM restore instance type that matches a VM restore instance type identified during performance of the search;
using the test VM as a basis to create a test restore VM at a cloud storage site;
restoring the test VM at the cloud storage site using the test restore VM;
generating a 4-D baseline vector based on the restoration of the test VM, the 4-D baseline vector identifying a particular VM restore instance type;
generating a 5-D vector based on the 4-D baseline vector;
ranking the 5-D vector relative to one or more other 5-D vectors, all of the 5-D vectors identifying the same production site VM; and
restoring, at the cloud storage site, the production site VM identified in the 5-D vectors, and the production site VM restored at the cloud storage site has a VM restore instance type identified in the highest ranked 5-D vector.