Browse Prior Art Database

Method to Provide Advice to a Storage Subsystem to Enable Intelligent Optimization Disclosure Number: IPCOM000199580D
Publication Date: 2010-Sep-09
Document File: 3 page(s) / 34K

Publishing Venue

The Prior Art Database


The invention disclosed relates to a method to provide advice to a storage subsystem to enable intelligent optimizations. A hypervisor (or in the absence of a hypervisor, the operating system in the application server) advises storage systems as to which set of storage devices (e.g. LUNs) can be expected to deliver I/O service and which ones will not be expected to deliver service (and optionally, for how long). Given this information, the storage subsystem can make intelligent decisions about optimizing resource usage in the subsystem (e.g. minimizing energy usage, maximizing use of critical resources to deliver best performance).

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 51% of the total text.

Page 1 of 3

Method to Provide Advice to a Storage Subsystem to Enable Intelligent Optimization

Today's disk storage subsystems generally have very little knowledge of the state of the applications running in application servers. Therefore, they must be ready to provide service, assuming that any application may require service at the highest performance level at any moment. This can cause the disk storage subsystem to provide service at significantly higher cost than necessary. For example, if service was not required by the applications for some period of time and this fact is known by the storage subsystem, certain optimizations might be made such as powering drives off or ensuring that the data wasn't stored in very high performance high cost disk drives or SSDs.

Today's application servers often include a virtual machine (VM) system consisting of a hypervisor which manages a set of guest virtual machines, each of which runs an application. The hypervisor allocates CPU resource (e.g. MIPS, memory, IO bandwidth) to a subset of the configured guest virtual machines according to demand along with configured allowed usage of the resources by each guest virtual machine. Therefore, the hypervisor is in the position of knowing which guest virtual machines are active at any point and which storage devices (e.g. LUNs) they are configured to use.

In such a system comprising one or more application servers (optionally) containing hypervisors and guest virtual machines, along with one or more storage subsystems containing a processing unit for running firmware, disk drives and solid state disk drives, this invention is a set of steps for providing this beneficial advice to the storage subsystem.


[This page contains 1 picture or other non-text object]

Page 2 of 3

Telling a storage subsystem that a LUN will not be used:

     The hypervisor determines that a VM is to be swapped out. It makes an estimate of the amount of time that a VM will be inactive. The hypervisor determines which Logical Unit Numbers (LUNs) are configured to that virtual machine (or perhaps set of virtual machines acting as a cluster) and which of those LUNs will not be needed by any other virtual machine or hypervisor process (such as backup). The hypervisor uses a communication channel with the storage subsystem to advise the storage subsystem that various of its LUNs will not be used, optionally with a period of time that the LUNs are not expected to be used. The storage subsystem updates a data structure indicating which of the LUNs are expected (or not) to be used.

     The storage system performs an analysis to determine optimization...