International Business Machines Corporation

James Venuto

David Alan Christenson

John Maxwell Cohn

Paul Steven Zuchowski

Ivan L Wemple

Stuart Irwin Feldman

Geoffrey D McNiven

Ronald Marz

Techniques are described for load balancing between DHCP servers at a DHCP client. Embodiments include receiving, at the DHCP client, a respective DHCPOFFER message from each of two or more DHCP servers, wherein each of the DHCPOFFER messages includes an IP address offer and load information describing a current workload of the respective DHCP server. One of DHCP servers is selected, at the DHCP client, to accept an offer for an IP address from, based on the received load information and a subnet on which the selected DHCP server is located. Embodiments transmit a DHCPREQUEST message, from the DHCP client, to the selected DHCP server, to accept the IP address offer from the selected DHCP server.

Client-driven load balancing of dynamic IP address allocation

Techniques are described for detecting duplicate IP addresses at a DHCP relay agent. Embodiments identify a list of DHCP clients that have valid IP address leases from one or more DHCP servers. Additionally, network traffic from a first node passing through the DHCP relay agent is monitored. Embodiments then determine whether the first node is using a duplicate IP address, based on the monitored network traffic and the identified list of DHCP clients.

Duplicate IP address detection by a DHCP relay agent

Techniques are described for load balancing between DHCP servers at a DHCP client. Embodiments receive load information from each of two or more DHCP servers. Here, the load information describes a current workload of the respective DHCP server from which the load information is received. One of the two or more DHCP servers is then selected at the DHCP client based on the received load information. Additionally, embodiments accept an IP address offer from the selected DHCP server.

Techniques are described for providing a backup DHCP server using a DHCP relay agent. Embodiments retrieve DHCP server configuration information specifying a configuration of one or more DHCP servers. Additionally, embodiments monitor, at a DHCP relay agent, DHCP lease traffic between the one or more DHCP servers and one or more DHCP clients passing through the DHCP relay agent. Upon determining that the one or more DHCP servers have failed, DHCP server services are provided at the DHCP relay agent, based on the received DHCP server configuration information and the monitored DHCP lease traffic.

Extending a DHCP relay to backup a DHCP server

A method of analyzing supply voltage drops in a power grid for distributing power to an integrated circuit chip during design. The method initially comprises providing a library of circuits for use in designing an integrated circuit chip and determining a supply current requirement and an operating voltage range for each circuit in the circuit library. The method then includes calculating an admittance matrix representing the power grid with a pre-specified array of circuit ports defined by intersection of the power grid and a modeling grid, assigning regions of the power grid to each of the ports, and placing a set of circuits from the circuit library in regions on the power grid. The method further includes calculating a total node current at each of the ports by summing current requirements of all of the circuits located in the regions, calculating a node voltage at each of the ports by solving a system of linear equations corresponding to the calculated admittance matrix, imposing a penalty to each node having a node voltage outside of a predetermined range, and calculating the node voltages and the penalties to a cost-based floorplanning/placement analysis tool.

Method for supply voltage drop analysis during placement phase of chip design

A method and apparatus for extending the main memory of computer systems across network boundaries is provided. A memory server is allocated on a network server, or a separate machine connected to a network to enable client computers on the network to extend their main memory on demand for user applications. Through the utilization of the memory server, the client workstations can execute programs requiring more memory than is resident on the client without decreasing the efficiency of the network server.

Growing community of inventors

Poughkeepsie, NY, United States of America

James Venuto