International Business Machines Corporation

International Business Corporation

Leonidas Georgiadis

Roch A Guerin

Israel Cidon

Yuval Y Shavitt

Andrew E Slater

Subir Varma

Vinod Gerard Peris

Christos N Nikolaou

Donald Francis Ferguson

Rajendran Rajan

George W Wang

The present invention addresses the issue of controlling delay variations (jitter) in packet-switched networks by enhancing the capabilities of existing scheduling policies. The idea is to use a few bits in the header of packets to send jitter control information to the downstream network elements, so that the delay variation caused by the upstream network element, is compensated for by the downstream network element. The key point to observe is that packets (or cells in ATM) may be small, and therefore, not contain many bits in the header that can be used for jitter control. We describe a unique scheme, that utilizes the bits that are available for jitter control in an efficient manner, allowing for the desired jitter to be obtained with as few bits as possible.

Method to control jitter in high-speed packet-switched networks

The present invention relates to the issue of providing end-to-end delay guarantees in a multi-node communication system. More specifically, the present invention addresses the problem of specifying operational parameters of rate-controlled service disciplines in a communication network in order to efficiently provide end-to-end delay guarantees. The key contribution is a method for specifying leaky bucket parameters as well as scheduling delays at each node, which are used as inputs to the rate-controlled service discipline.

Traffic flow regulation to guarantee end-to-end delay in packet switched

A non-quota access indicator is circulated among nodes in a multi-node quota based communication system with a shared resource, indicating maximum possible non-quota access to the shared resource to a given node receiving same. Upon arrival at a node, the indicator is saved and then updated to reflect the current status of that node as either starved or satisfied, the former being a condition of currently having quota remaining and a shared resource access requirement, and the latter being a condition of either currently having no remaining quota or having no current shared resource access requirement. After updating, the node immediately propagates the indicator to the next node in the system. When a node without quota requires access to the shared resource, it compares its requirement to the last stored indicator and accesses the shared resource if the stored indicator is equal to or greater than the access requirement.

Method and apparatus for improved throughput in a multi-node

In a multiple processor transaction processing system, a class response time goal for each class of transactions is enforced by a workload manager, which computes a current average class response time for each transaction class and derives a current class performance index for each of these classes with respect to its class response time goal. Whenever a transaction arrives, the workload manager considers a number of different possible transaction servers to which that arriving transaction could be routed and predicts estimated new values for the class performance indices for each of the considered routing choices. An overall goal satisfaction index is determined for each one and the routing choice corresponding to the best overall goal satisfaction index is selected as the routing choice. In the preferred embodiment, the workload manager also priority rates or orders the classes in accordance with the current class performance indices so that a transaction of a class which is performing more poorly (as judged by its class performance index) gets a higher dispatch priority at the server to which it is routed than a transaction of a class that is performing better.

Workload manager for achieving transaction class response time goals in

A counter for tracking current shared resource access quota requests of nodes in a multi-node system is circulated among the nodes. A quota for a given node is determined as a function of the counter, the needs of the given node and a variable indicative of total access quota for all nodes combined. The determined quota is then allocated to the given node, i.e., the node is allowed to access the shared resource in accordance with the determined quota.

Method and apparatus for dynamically allocating shared resource access

A semi-dynamic load balancer for a transaction processing system reallocates transaction types among computers in the system as a group rather than as individual transactions. Statistical data is accumulated in an affinity matrix which records the number of times a transaction type i was blocked by transaction type j was holding a lock on a data item that transaction type i wanted to access. When an overloaded computer is detected, transaction types are reallocated on the computers of the system according to a transaction type routing table which is updated based on data from the affinity matrix.

Growing community of inventors

Chappaqua, NY, United States of America

Leonidas Georgiadis