Hyper-V does not support memory overcommitment, and has been criticized by VMware and other virtualization vendors. Of course, Microsoft has been denied that this is their weakness, that in order to ensure the production environment VM performance, excessive allocation of memory is inappropriate. But Microsoft's attitude seems to have changed abruptly, March 18, in its Virtualization official blog published an article, claiming that Microsoft has listened to a lot of customers, think how to avoid performance degradation at the same time, the maximum utilization of resources and investment, is the customer's concern, but also the direction of Microsoft's efforts, So they plan to launch "dynamic memory" technology.
Http://blogs.technet.com/virtualization/archive/2010/03/18/dynamic-memory-coming-to-hyper-v.aspx
The specifics of the technical details are still unknown, but look at the description, much like VMware's memory overcommitment (excessive memory allocation).
Interestingly, however, Microsoft offers a sliding adjuster that gives users more choice. My understanding is that sliding to the left, the pursuit of the highest performance, is the equivalent of the current, completely not excessive allocation of memory. As to how these scales are designed in the middle, Microsoft has yet to provide more details.
It is not known whether a technology like TPS (transparent Page sharing) will be adopted by Microsoft.
In any case, this is an interesting change and will make Hyper-V more competitive.
Additional information was provided, supplemented by the following:
The Hyper-V feature, released in October 2008, contains dynamic Memory, but was cut off during the summer of 2009 when Hyper-V was released, perhaps at a time when the technology was immature.
Http://hypervoria.com/hyper-v/hyper-v-2-0-feature-overview.aspx
The article refers to the dynamic Memory, cited below:
Overview
* Pool of memory is dynamically distributed across VMs
* Memory is dynamically allocated/removed based VM usage with no service interruption
Benefits
* Enables much higher consolidation ratios each host by addressing the greatest limiting-factor to consolidation:memory
How
VM memory configuration includes:
* Initial (what VM would boot with)
* Minimum (what VM is guaranteed)
* Maximum (what VM can grow to)
* Memory is added via Hot-add MEM functionality
* Memory is removed via Balloon driver (supported OSs)
Seen here, minimum, similar to VMware's memory reservation,maximum, is similar to the memory Limit of VMware, and the memory between minimum and Maximum, Can be provisioned between VMS through the balloon drive ... This and VMware's memory implementation mechanism is very similar!