| 摘要 |
The present invention relates to a memory control method for a virtual system which controls memory assigned to multiple virtual machines in an asymmetric memory access architecture which has multiple nodes comprising: a core running each process, a common cache, a memory controller, an interconnection, and a memory. The method may include the steps of: by each node, extracting page frames used as buffer cache among the page frames assigned to the multiple guest operating systems of the multiple virtual machines, and dividing the page frames according to patterns with respect to data access into a sequential pattern, a loop pattern, and a non-classified pattern; by each node, managing the divided page frames according to LRU policy, wherein the page frames belonging to the sequential pattern and the loop pattern are managed by a continuous frame unit, and the page frames belonging to the non-classified pattern are managed by a page frame unit; and reassignment of memory between the multiple virtual machines, taking into account at least one of interconnection latency for each node. |
