Frederick B Kiremidjian

US Patent:

20030081623, May 1, 2003

Filed:

Oct 27, 2001

Appl. No.:

10/004078

Inventors:

Frederick Kiremidjian - Danville CA, US
Li-Ho Hou - Saratoga CA, US

Assignee:

AMPLIFY.NET, INC.

International Classification:

H04L012/28

US Classification:

370/412000

Abstract:

A method comprises using a class-based queue traffic shaper that enforces multiple service-level agreement policies on individual connection sessions by limiting the maximum data throughput for each connection. The class-based queue traffic shaper distinguishes amongst datapackets according to their respective source and/or destination IP-addresses. Each of the service-level agreement policies maintains a statistic that tracks how many datapackets are being buffered at any one instant. A test is made of each policy's statistic for each newly arriving datapacket. If the policy associated with the datapacket's destination is currently buffering, or holding, any datapackets, then the newly arriving datapacket is sent to be buffered too. This allows the longest waiting datapacket for the particular destination to be released and cleared from the buffer first.

US Patent:

20030099198, May 29, 2003

Filed:

Nov 27, 2001

Appl. No.:

10/004079

Inventors:

Frederick Kiremidjian - Danville CA, US
Li-Ho Raymond Hou - Saratoga CA, US

Assignee:

AMPLIFY.NET, INC.

International Classification:

H04L012/56

US Classification:

370/230100, 370/390000, 370/412000

Abstract:

A multicast method generates a packet-tracking queue with individual entries representing datapackets for transfer through a hierarchical network. The actual datapacket and its real payload are stored as one item in a FIFO buffer. Any packet-tracking queue entry representing a multicast datapacket is expanded into several consecutive entries, one each for the individual subscribers enrolled to receive a broadcast. A first of such expanded entries is flagged as being first, and a last such expanded entry is flagged as being last. Each expanded entry may be subject to its own unique service-level policy, with the result that the datapackets can be released and cleared in any order. If the first or last are released, then the next are marked as first or last. When only one expanded entry is left, it will be flagged as being both first and last. When it is released, the entry in the packet-tracking queue is cleared as well as the actual datapacket and its real payload in the FIFO buffer. Such individually delayed releases of the multicast datapackets help enforce the bandwidth limiting function of the service-level policies involved.

US Patent:

20030099199, May 29, 2003

Filed:

Nov 27, 2001

Appl. No.:

10/004080

Inventors:

Frederick Kiremidjian - Danville CA, US
Li-Ho Raymond Hou - Saratoga CA, US

Assignee:

AMPLIFY.NET, INC.,

International Classification:

H04L012/56

US Classification:

370/230100, 370/390000, 370/412000

Abstract:

A network-node bandwidth-allocation credit method includes computing credits after each completed scan of a packet-tracking queue. Such queue varies tremendously in depth, according to how much network traffic is transitioning through the involved network nodes. A bandwidth traffic-shaping manager operates to control the maximum bandwidth permitted to pass through each network node, e.g., by buffering datapackets that would exceed some service policy limit if forwarded immediately on receipt. As each network node runs less that its policy maximum, it is given a number of credits that collect in a bank account. If a datapacket presents itself that involves passage through the network node, such bank account is checked to see if sufficient bandwidth-allocation credits exist to forward the datapacket immediately. If so, an appropriate deduction of credits is made and the datapacket is forwarded toward its destination.

US Patent:

20030099200, May 29, 2003

Filed:

Nov 27, 2001

Appl. No.:

10/004608

Inventors:

Frederick Kiremidjian - Danville CA, US
Li-Ho Hou - Saratoga CA, US

Assignee:

AMPLIFY.NET, INC.

International Classification:

H04L012/26

US Classification:

370/230100, 370/412000

Abstract:

A method and a semiconductor intellectual property embodiment of the present invention comprise a class-based queue traffic shaper that enforces multiple service-level agreement policies on individual connection sessions by limiting the maximum data throughput for each connection. The class-based queue traffic shaper distinguishes amongst datapackets according to their respective source and/or destination IP-addresses. All limit checking is done in one clock cycle for the entire network hierarchy above a particular node, and previously independent and separate queues are combined into one super queue.

US Patent:

20030229714, Dec 11, 2003

Filed:

Jun 5, 2002

Appl. No.:

10/163408

Inventors:

Frederick Kiremidjian - Danville CA, US
Li-Ho Hou - Saratoga CA, US

Assignee:

Amplify.Net, Inc.

International Classification:

G06F015/173

US Classification:

709/244000

Abstract:

A semiconductor integrated circuit chip comprises a class-based queue traffic shaper that enforces multiple service-level agreement policies on individual connection sessions by limiting the maximum data throughput for each connection. The class-based queue traffic shaper distinguishes amongst datapackets according to their respective source and/or destination IP-addresses. Each of the service-level agreement policies maintains a statistic that tracks how many datapackets are being buffered at any one instant. A test is made of each policy's statistic for each newly arriving datapacket. If the policy associated with the datapacket's destination indicates the agreed bandwidth limit has been reached, the datapacket is buffered and forwarded later when the bandwidth would not be exceeded.

US Patent:

20030229720, Dec 11, 2003

Filed:

Jun 5, 2002

Appl. No.:

10/163407

Inventors:

Frederick Kiremidjian - Danville CA, US
Li-Ho Hou - Saratoga CA, US

Assignee:

Amplify. Net, Inc.

International Classification:

G06F015/173
G06F015/16
G06F015/177

US Classification:

709/253000, 709/232000

Abstract:

A heterogeneous-network switch comprises a number of different-type media access controllers (MAC's) each respectively for connection to otherwise incompatible computer networks. An incoming data bus is connected to collect datapackets from each of the different-type MAC's. An outgoing data bus is connected to distribute datapackets to each of the different-type MAC's. And, a traffic shaping cell (TSCELL) having an input connected to the incoming data bus and an output connected to the outgoing data bus, provides for traffic control of said datapackets according to a bandwidth capacity limit of a corresponding one of said otherwise incompatible computer networks to receive them. The switch is based on a class-based queue traffic shaper that enforces multiple service-level agreement policies on individual connection sessions by limiting the maximum data throughput for each connection. The class-based queue traffic shaper distinguishes amongst datapackets according to their respective source and/or destination IP-addresses. Each of the service-level agreement policies maintains a statistic that tracks how many datapackets are being buffered at any one instant. A test is made of each policy's statistic for each newly arriving datapacket. If the policy associated with the datapacket's destination indicates the agreed bandwidth limit has been reached, the datapacket is buffered and forwarded later when the bandwidth would not be exceeded.

US Patent:

47274751, Feb 23, 1988

Filed:

May 18, 1984

Appl. No.:

6/611763

Inventors:

Frederick Kiremidjian - Dublin CA

International Classification:

G06F 1516
G06F 1322

US Classification:

364200

Abstract:

An initial polling sequence for configuring a modular computer system employing a system bus for interconnecting the CPU and various modules attached to the bus. At the beginning of the polling sequence, a bus base address register in each module is preset to a port 0 address by the CPU. A POLL signal is generated by the CPU and intercepted by the closest module, which responds by placing a module identification character on the data lines of the system bus. The CPU receives the module identification character, stores this character in a table and issues a bus base address for that module. Upon receipt of the bus base address, the module presently active in the polling sequence issues a POLL command to the next module on the system bus. The poll sequence is repeated until all modules have been assigned and have received a bus base address.

US Patent:

47046970, Nov 3, 1987

Filed:

Jun 17, 1985

Appl. No.:

6/745621

Inventors:

Frederick B. Kiremidjian - Dublin CA
Jefferson C. Buchanan - San Jose CA

Assignee:

Counterpoint Computers - San Jose CA

International Classification:

G06K 300
G06F 1540

US Classification:

364518

Abstract:

There is described a memory system for use in a data processing system having more than one bit mapped display station. The memory system uses a single memory array for storing at least one separate full screen of bit mapped video data for each display station. A single video controller controls the process by which video data is read from the memory array and then simultaneously transmitted to each of the display stations. Furthermore, a memory access control unit coordinates use of the memory by the video controller and the data processing unit and also translates address signals generated by the data processing unit so that it can access the video data for each display station using a contiguous set of address values.