Since a typical backbone links have

Since a typical backbone links have more than 20,000 flows, those flows from different hosts are not synchronized. Thus the buffer size requirement can be lowered because the flow and con- gestion control mechanisms will not simultaneously require a large buffer for each flow. Therefore the buffer size in an interface can be dramatically reduced to save cost and improve performance.
In recent publications [24][25] the buffering needed for N TCP flows is derived as which is referred as the small buffer model.

Результаты (английский) 2:

Since a typical backbone links have more than 20,000 flows, those flows from different hosts are not synchronized. Thus the buffer size requirement can be lowered because the flow and con- gestion control mechanisms will not simultaneously require a large buffer for each flow. The buffer 'size bed Therefore in an the interface CAN the BE Dramatically reduced A to the save cost and Improve performance.
With In recent publications [24] [25] the buffering needed for N TCP flows is derived as which is referred as the small buffer model.

Результаты (английский) 3:

since a typical backbone links have more than 20000 flows, those flows from different hosts are not synchronized. thus the buffer size requirement can be lowered because the flow and con - gestion control mechanisms will not simultaneously for a large buffer for each flow. therefore the buffer size in an interface can be dramatically reduced to save cost and improve performance.in recent publications [24] [25] the buffering needed for n tcp flows is derived, which is referred as the small buffer model.