A hierarchical structure comprising a physical port bandwidth that is divided into a plurality of logical links, the bandwidth available to each of the logical links is divided into plurality of VLANs, and the bandwidth associated with each VLAN is shared by a plurality of individual user flows;

A plurality of SRAM and DRAM memory devices external to the scheduler, a first SRAM that contains a set of data buses and that stores flow queue control blocks that are required for ‘read’ and ‘write’ at each flow queue enqueue time, and a second SRAM containing two data buses, one dedicated to ‘read’ and one dedicated to ‘write’, and that stores frame control blocks and VLAN hierarchy control blocks, a first DRAM that stores a network management counter and a second DRAM that stores flow queue control blocks and VLAN hierarchy control blocks that require ‘read’ only at both enqueue time and dequeue time;

Control blocks of scheduling elements stored in said SRAM and DRAM memory devices with at least some of the memory devices storing more than one type of control block; wherein

a) SRAM memory is used if the content of a control block is Read-Modify-Write at packet enqueue and at dequeue;

b) SRAM and DRAM memory are used if the control block content is Read-Modify-Write only, at the packet dequeue;

c) DRAM memory is used if the control block content is Read only at packet enqueue and dequeue; and

d) Memory access allocated to enqueue tasks does not conflict with memory access allocated to dequeue tasks;

Time based calendar arrays to provide guaranteed bandwidth service for flow queues and VLANs; and

Weighted fair queueing calendar arrays for allocation of available bandwidth for competing flows when no service is required by time base calendars.