Efficiency of virtual machines that use de-duplication as primary data storage
View Patent ↗Example apparatus and methods provide two types of storage for a virtual machine running on a hypervisor. The first storage is de-duplication based and the second storage is not de-duplication based. Example apparatus and methods may acquire data from the first storage to instantiate the virtual machine, to instantiate an operating system on the virtual machine, or to instantiate an application on the virtual machine from the first storage. Example apparatus and methods may write a snapshot to the second storage and then support random input/output for the virtual machine, for the operating system, or for the application from the second storage. The snapshot may selectively be collapsed or the second storage may selectively be retired and thus example systems may selectively update the first storage from the second storage. Having dual devices facilitates using de-duplication storage for de-duplication-centric I/O while non-de-duplication storage is used for random I/O.
1. An apparatus, comprising:
a processor configured to detect an input/output operation associated with a virtual machine running on a hypervisor;
a first interface to a de-duplication storage device that stores de-duplicated data associated with the virtual machine;
a second interface to a first non-de-duplication storage device that stores a first snapshot associated with the virtual machine;
an identification logic configured to select a target device from which the input/output operation should be satisfied, the target device being selected from the de-duplication storage device and the first non-de-duplication storage device, the target device being selected as a function of an efficiency condition; and
an operation logic configured to satisfy the input/output operation from the target device.
2. The apparatus of claim 1 , the input/output operation originating in an operating system running on the virtual machine or in an application running on the virtual machine.
3. The apparatus of claim 2 , the identification logic being configured to select the de-duplication storage device when the input/output operation is associated with instantiating the virtual machine, instantiating the operating system, or instantiating the application.
4. The apparatus of claim 1 , where the de-duplication storage device is part of the apparatus or the first non-de-duplication storage device is part of the apparatus.
5. The apparatus of claim 1 , comprising a transfer logic configured to control selectively transferring data from the first non-de-duplication storage device to the de-duplication storage device upon occurrence of a transfer event.
6. The apparatus of claim 5 , the transfer event being the first snapshot reaching a fullness threshold, the first non-de-duplication storage device reaching a fullness threshold, the first non-de-duplication storage device reaching an idleness threshold, detecting termination of the virtual machine, detecting termination of an operating system running on the virtual machine, or detecting termination of an application running on the virtual machine.
7. The apparatus of claim 5 , the transfer logic being configured:
to establish a second, different snapshot on a second different non-de-duplication storage device before updating the de-duplication storage device from the first non-de-duplication storage device;
to reconfigure the identification logic to select the target device from the de-duplication storage device and the second non-de-duplication storage device before updating the de-duplication storage device from the first non-de-duplication storage device; and
removing the first snapshot from the first non-de-duplication storage device after updating the de-duplication device from the first non-de-duplication storage device.
8. The apparatus of claim 1 , the efficiency condition being identifying the input/output operation as being associated with booting the virtual machine, identifying the input/output operation as being associated with instantiating an operating system on the virtual machine, identifying the input/output operation as being associated with instantiating an application on the virtual machine, a capacity of the first non-de-duplication storage device, a capacity of the first snapshot, a predicted time to complete the input/output operation from the de-duplication storage device, a predicted time to complete the input/output operation from the first non-de-duplication storage device, a measure of how busy the de-duplication storage device is, or a measure of how busy the first non-de-duplication storage device is.
9. The apparatus of claim 1 , the efficiency condition being identifying which of the de-duplication storage device and the non-de-duplication storage device can serve the input/output operation more efficiently.
10. The apparatus of claim 1 , the identification logic being configured to select the first non-de-duplication storage device when the input/output operation is associated with a non-instantiating operation performed on the virtual machine.
11. A non-transitory computer-readable medium storing computer-executable instructions that when executed by a computer cause the computer to perform a method, the method comprising:
upon detecting an input/output operation associated with a virtual machine running on a hypervisor:
instantiating the virtual machine with data read from a first storage apparatus, the first storage apparatus being configured to store data using data de-duplication;
selecting a second storage apparatus from which a data storage operation using data de-duplication should be satisfied, the second storage apparatus being selected from the first storage apparatus and a non-de-duplication storage device, the second storage apparatus being selected as a function of an efficiency condition;
producing a first snapshot of the virtual machine in the second storage apparatus, the second storage apparatus being configured to store data without using data de-duplication;
upon detecting another input/output operation that can be satisfied from the second storage apparatus, the other input/output operation being associated with the virtual machine, satisfying the other input/output operation from the second storage apparatus;
upon detecting that a transfer condition exists for the second storage apparatus, updating the first storage apparatus from the second storage apparatus, where the transfer condition is a fullness level of the first snapshot, a fullness level of the second storage apparatus, termination of the virtual machine, termination of an operating system running on the virtual machine, termination of an application running on the virtual machine, or an idleness level of the second storage apparatus;
where updating the first storage apparatus from the second storage apparatus includes producing a second snapshot before updating the first storage apparatus from the second storage apparatus, where producing the second snapshot includes writing at least a portion of virtual memory associated with the virtual machine to a third storage apparatus, the third storage apparatus being configured to store data without using data de-duplication; and
where updating the first storage apparatus from the second storage apparatus includes transferring segments of data from the second storage apparatus to the first storage apparatus, where the segments are larger than segments associated with a file system associated with the virtual machine, and where the segments are transferred using sparse-aware sequential input/output.
12. The non-transitory computer-readable medium of claim 11 , where instantiating the virtual machine includes reading blocks of de-duplicated data from the first storage apparatus using sequential input/output.
13. The non-transitory computer-readable medium of claim 11 , including instantiating an operating system or an application to run on the virtual machine by reading blocks of de-duplicated data from the first storage apparatus using sequential input/output.
14. The non-transitory computer-readable medium of claim 11 , where producing the first snapshot includes writing at least a portion of virtual memory associated with the virtual machine to the second storage apparatus.
15. The non-transitory computer-readable medium of claim 11 , where detecting the input/output operation that can be satisfied includes identifying a write that will fit in the second storage apparatus, identifying a read for which data is present in the first snapshot or in the second storage apparatus, or identifying a write that will overwrite data already written to the second storage apparatus.
16. The non-transitory computer-readable medium of claim 11 , where satisfying the other input/output operation from the second storage apparatus includes reading data from the first snapshot.