<![CDATA[Consider the packets switch node's buffer, which is shared by numerous output communications lines. Spreading buffered memory across numerous users minimizes the total amount of storage required to fulfill latencies constraints and the possibility of packets loss. despite this, there can be an issue with assigning buffers memory amongst various users given that specific users who have consumed most overall that memory could limit or restrict accessibility to communications connections for others users, considerably reducing the overall efficiency of the switch node. These are several buffers storage allocation strategies, among which, known as SMA (Share in Minimum Assignment), are being studied in this research to decrease the expenses related to packets denial and postponement, as well as the functioning of the drives and lines of communication. The switch nodes are modelled using a multithreaded queue system with parallel devices of the kind, a memory buffer sharing accordingly to the SMA scheming, and a set amount of memory spaces designated to every device. A mathematical description of the problems of optimising the SMA schemes in regard to the amount of publically available buffer placements is presented with the goal to reduce losses to the system caused by applications disapproval, application queuing latency, and buffers and device operation. The hypothesis about the bounds of the field that contains the global optimal point is proven. A variety of arguments are also provided as a result of the theorem regarding the location of the globally optimal of the function of objective for different switch nodes types and specific instances of SMA.]]>
Copyrights © 2025
