International Business Machines Corporation

Lenovo (Singapore) Pte. Ltd.

International Business Corporation

Jorge Rolando Rodriguez

Ganesh Balakrishnan

Kaiqi Xiong

Shailabh Nagar

Metin Aydemir

Darryl Edward Gardner

Andrew J Rindos, Iii

Ricardo Paez

James Milton Stephens, Jr

Embodiments of the present invention provide a method, system and computer program product for multi-scale network traffic generation. In one embodiment of the invention, a network traffic generation method can be provided. The method can include defining multiple, different scales in an n-state MMPP model to accommodate a full characteristic response of a modeled traffic scenario. The method further can include establishing a transition window for each of the scales and determining a state through the transition window for selected ones of the scales. Finally, the method can include computing an inter-packet time according to the determined state for each of the selected ones of the scales and generating and transmitting packets for the selected ones of the scales utilizing a correspondingly computed inter-packet time.

Multi-scale network traffic generator

A design structure embodied in a machine readable storage medium for at least one of designing, manufacturing, and testing a design is provided. The design structure includes a network traffic generation system. The system can include a Markov modified Poisson process (MMPP) model, a packet scheduler coupled to the MMP model, a data store of transition windows defined for different defined scales, traffic generation parameter computing logic comprising program code enabled to compute traffic generation parameters for different scales according to respective states identified within different transition windows in the data store for the different scales, and a packet transmitter coupled to the packet scheduler.

Structure for a multi-scale network traffic generator

Methods and products are disclosed for analyzing network traffic using an improved Markov Modulated Poisson Process Model with one bursty state and a plurality of idle states that include: establishing a time scale of operation for each state in the improved MMPP model; establishing a transition value for each state in dependence upon the time scale of operation for the state; measuring inter-arrival times between individual packets received in one or more network adapters; and determining a current state for the network traffic independence upon the measured inter-arrival time of a most recently received packet and the transition values.

Analyzing and generating network traffic using an improved Markov Modulated Poisson Process model with one bursty state and a plurality of idle states

Data communication in network traffic is modeled in real time and is analyzed using a 2-state Markov modified Poissen process (MMPP). The traffic inter-arrival times for bursty and idle states define a transition window [&#x3bb;, &#x3bb;] represented by the boundary values &#x3bb;for the inter-arrival time for bursty traffic, and &#x3bb;for the inter-arrival time for idle traffic. Changes in the values of &#x3bb;and &#x3bb;are tracked over time, and the size of the transition window is enlarged or decreased based upon relative changes in these values. If the inter-rival times for the bursty state and the idle state become approximately equal, the model defaults to a single state model. The modeling is applicable to the synchronization of polling and blocking in a low-latency network system. This permits the adoptive selection of poll or block to maximize CPU utilization and interrupt latency.

MMPP analysis of network traffic using a transition window

Aspects of the invention provide an improved solution for processing packets in a packetized communications network. For example, a next packet in a set of incoming packets placed in a plurality of queues is selected by obtaining a random/pseudo-random search key and identifying one of the plurality of queues based on the search key and a Patricia tree that includes at least one child node for each of the plurality of queues. A greedy algorithm can be used to select an alternative queue should the first selected queue be empty.

Multi-queue packet processing using Patricia tree

Methods, apparatus, and products are disclosed for analyzing network traffic using an improved Markov Modulated Poisson Process Model with two barrier states that include: retrieving a previous state for the network traffic; measuring inter-arrival times between individual packets received in one or more network adapters; establishing a transition window in dependence upon the measured inter-arrival times, the transition window having a transition value &#x3bb;that represents an upper boundary for the inter-arrival times in a bursty state and having a transition value &#x3bb;that represents a lower boundary for the inter-arrival times in an idle state; retrieving a previous fence value that prevents premature transitions into the idle state or the bursty state; and determining a current state for the network traffic in dependence upon the previous state for the network traffic, an inter-arrival time of a most recently received packet, the transition values, and the previous fence value.

Analyzing network traffic using an improved markov modulated poisson process model with two barrier states

Self-similar data communication in network traffic is modeled real time and is analyzed using a Markov modified Poissen process (MMPP) to characterize the traffic flow and to accommodate high variability in traffic flow from one time period to the other. The analysis is performed at multiple time levels using a bottom-up approach. The parameters of the model are adjustable at each level according to the traffic parameters at that level. Each model consists of 2 states of network traffic behavior comprising a bursty state representing heavy traffic conditions and an idle state representing light traffic conditions. A transition window defines the upper time interval for the receipt of packets in the bursty state and the lower time interval for the receipt of packets in the idle state. If the inter-rival times for the bursty state and the idle state become approximately equal, the model defaults to a single state model.

Multilevel analysis of self-similar network traffic

A method that avoids cache flooding when handling sequential I/O request streams in storage subsystem caches. Both on-demand and prefetched data are addressed. A mechanism is provided that makes the blocks read from the disk array, that are part of a sequential I/O request stream, minimize the evictions of potentially useful data blocks from the cache, thus avoiding the flooding of the cache by stale data.

Preventing cache floods from sequential streams

A system, computer program product and method for reallocating memory space for storing a partitioned cache. A server may be configured to receive requests to access a particular logical drive. One or more logical drives may be coupled to an adapter. A plurality of adapters may be coupled to the server. Each logical drive may be associated with one or more stacks where each stack may comprise one or more cache entries for storing information. The one or more stacks associated with a logical drive may be logically grouped into a logically grouped stack associated with that logical drive. Each of the logically grouped stacks of the one or more logical drives coupled to an adapter may be logically grouped into a logically grouped stack associated with that adapter. By logically grouping stacks, memory supporting a partitioned cache may adaptively be reallocated in response to multiple criteria thereby improving the performance of the cache.

Partitioned cache of multiple logical levels with adaptive reconfiguration based on multiple criteria

A system, computer program product and method for designing a cache. A server in a network system, e.g., file system, database system, may receive requests forming a workload. A trace may be performed on the workload to provide information such as the frequency count for each Logical Block Address (LBA) requested in the workload. The trace may then be analyzed by grouping the LBA's with the same frequency count and determining the number of groups counted in the trace. Upon analyzing the trace, an LRU-LFU cache may be designed. An LRU-LFU cache may comprise one or more stacks of cache entries where the number of stacks corresponds to the number of frequency groups counted in the trace. Each particular stack may then have a length based on the number of logical addresses with the same frequency count associated with that particular stack.

Designing a cache using an LRU-LFU array

A system, computer program product and method for reconfiguring a cache. A cache array may be created with one or more stacks of cache entries based on a workload. The one or more stacks may be ordered from most frequently used to least frequently used. The cache entries in each particular stack may be ordered from most recently used to least recently used. When a cache hit occurs, the cache entry requested may be stored in the next higher level stack if the updated frequency count is associated with the next higher level stack. When a cache miss occurs, the cache entry in a least recently used stack position in the stack with the lowest number of cache hits in the one or more stack positions tracked during a particular period of time may be evicted thereby allowing the requested information to be stored in the lowest level stack.

Designing a cache with adaptive reconfiguration

A method and system for synchronizing message transfers from a sender to a receiver, such that message latency and overhead processing are minimized. A next inter-message arrival delay is initially predicted in accordance with traffic history. The predicted inter-message arrival delay is categorized as either an intra-burst delay or an inter-burst delay. In response to a prediction of an inter-burst delay, the receiver is operated in a blocking mode wherein the receiver waits for an interrupt request from the sender. In response to a prediction of an intra-burst delay, the receiver is switched to a polling mode wherein the receiver polls for a message complete signal from said sender.

Method and system for reducing latency in message passing systems

A system having a high speed switching interface between remote buses is disclosed herein. The system includes a switch fabric to which multiple remote bus interfaces are connected. A programmable flow queue that includes multiple parallel logic flow queues is utilized for scheduling packets in accordance with protocol requirements of the remote bus interface.

Flow architecture for remote high-speed interface application

A system, computer program product and method for servicing requests. A server may be configured to receive a stream of requests to access particular logical block addresses in one or more logical drives in a RAID from one or more clients. The server may be coupled to one or more RAID adapters that are coupled to the RAID. The server may comprise a software RAID and each RAID adapter may comprise a hardware RAID. By monitoring the utilization of the processors in the server and in each RAID adapter, all or part of these received requests may subsequently be routed to either the software RAID or the hardware RAID based on which implementation is more desirable to service these requests.

Growing community of inventors

Cary, NC, United States of America

Jorge Rolando Rodriguez