System and method for monitoring performance, analyzing capacity and utilization, and planning capacity for networks and intelligent, network connected processes

• US 6,885,641 B1
• Filed: 12/01/1999
• Issued: 04/26/2005
• Est. Priority Date: 03/12/1999
• Status: Expired due to Term

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Chan, Ken, Silverman, Robert M., Klassen, Fredrick K. P.
• Primary Examiner(s): Nguyen, Chau
• Assistant Examiner(s): Duong, Duc
• Application Number:
  US09/452,403
• Time in Patent Office: 1,973 Days
• Field of Search: 370/252, 370/230, 370/230.1, 370/231, 370/232, 370/233, 370/234, 370/235, 370/250, 370/253, 370/248, 370/249, 709/223, 709/224, 709/225, 709/207, 709/237, 709/238, 709/240
• US Class Current: 370/252
• CPC Class Codes: H04L 41/142 using statistical or mathematical methods
• CPC Class Codes: H04L 43/0852 Delays
• CPC Class Codes: H04L 43/0882 Utilisation of link capacity
• CPC Class Codes: H04L 43/0888 Throughput
• CPC Class Codes: H04L 43/50 Testing arrangements

Abstract

Analysis of networks and testing and analyzing intelligent, network connected devices. An instantaneous network utilization value is assigned for the worst surviving ping instance of between 90% and 99% (determined proportionately from the ratio of dropped test samples to surviving test samples), and then used to solve for average network message size and average utilization of the network. A plurality transactions of different types are transmitted across the network to intelligent end systems and the results mathematically evaluated to determine the portion of the total response time contributed by the network and by the end processors; the utilization of the end processor processing subsystems and of the end processor I/O subsystems; and the utilization of the end system as a whole; and of the network and end processors considered as a unitary entity. Steps include determining utilization of the network when test packets are dropped by the network; utilization of intelligent processor and other devices attached to the network when test transactions are dropped, and when not dropped; and response time for remote processes at both the network and processor level.

Claims

1. A system for monitoring performance, capacity, and utilization of a network system, comprising:

means for sending probative test packets and transactions including short pings and long pings across a network to an intelligent end station to generate test results including best observed short ping and best observed long ping;

means for deriving and applying network queue factor to said test results to determine capacity, utilization, and performance of said network, devices connected by said network, and said devices and said network considered as a unitary system;

said test results further comprising data for instantaneously analyzing the current condition of said network system, establishing an historical database, determining how a user application transaction performs, and providing what if analysis for future changes to said network system.

2. A program storage device readable by a machine, tangibly embodying a program of instructions executable by a machine to perform method steps for monitoring performance, capacity, and utilization of a network system, said method steps comprising:

sending probative test packets and transactions across a network to an intelligent end station;

responsive to said test packets, calculating latency and average queue delay;

calculating apparent and streaming bandwidths;

calculating network hop count, multi-server factor, and queue depth; and

therefrom

determining capacity, utilization, and performance of said network, devices connected by said network, and said devices and said network considered as a unitary system.

3. The program storage device of claim 2, said method steps further comprising:

deriving and applying network queue factor to said test results to analyze the current condition of said network system;

establishing an historical database of said test results;

responsive to said test results, analyzing future changes to said network system.

4. The program storage device of claim 2, said sending step further comprising transmitting a plurality transactions of different types across the network to intelligent end systems;

and said determining step further comprising the steps of;

calculating the portion of total response time contributed by said network;

calculating the portion of said total response time contributed by end processors;

calculating the utilization by processing subsystems of said end processor;

calculating the utilization by input/output systems of said processors;

calculating the utilization of said intelligent end system as a whole; and

calculating the utilization of said network system, including said network and said intelligent end processors considered as a unitary entity.

5. System for monitoring performance, capacity, and utilization of a network system, comprising:

means for transmitting through said network system pings of different sizes to generate test data; and

mean for determining apparent responsiveness of said network as a function of latency and average queue delay, apparent and streaming bandwidths, network hop count, multi-server factor, and queue depth responsive to said test data.

6. The system of claim 5, said apparent responsiveness including a measure of network performance, processor performance, and network and processor performance considered as a unitary whole.

7. Method for monitoring performance, capacity, and utilization of a network system, comprising the steps of:

transmitting through said network system pings of different sizes to generate test data; and

responsive to said test data, calculating latency and average queue delay, apparent and streaming bandwidths, network hop count, multi-server factor, and queue depth for determining apparent responsiveness of said network system.

8. The method of claim 7, further comprising the steps of:

said transmitting step including performing long and short ping tests in isolation and in bursts to construct mathematical equivalents of isolation tests and burst tests.

9. Method of claim 7 for monitoring performance, capacity, and utilization of a network system, including the steps of:

generating network system test ping and transaction data;

performing mathematical analysis of said test ping and transaction data, including determining network end-to-end throughput and response time bandwidths, end processor transaction capacity, network end-to-end queue delay, network end-to-end latency delay, network internal packet size, network utilization, and utilization of intelligent processors connected by said network.

10. Method of claim 7 further for analyzing end-to-end bandwidth of said network system, comprising the steps of:

calculating queue depth; and

responsive to said queue depth,

determining processor utilization as a whole, and

determining bound operations utilization.

11. Method of claim 7 for monitoring performance, capacity, and utilization of a network system, comprising:

testing for presence of prioritization support within said network system; and

, if present,

measuring the capacity, utilization, and performance of said network system with respect to a plurality of priority levels by transmitting and analyzing sample packets and transactions set at varying priorities.

12. Method of claim 7, further for determining utilization of a system including a network which is dropping test packets, comprising the steps of:

assigning an instantaneous network utilization value for a worst surviving ping instance determined proportionately from a ratio of dropped test samples to surviving test samples; and

then

determining average network message size and average utilization of said network.

13. A method for monitoring performance, capacity, and utilization of a network system, comprising the steps of:

transmitting through said network system pings of different sizes to generate test data;

said transmitting step including performing long and short ping tests in isolation and in bursts to construct mathematical equivalents of isolation tests and burst tests;

responsive to isolation tests, calculating latency and average queue delay;

calculating apparent and streaming band widths;

calculating network hop count, multi-server factor, and queue depth; and

responsive to said test data, determining apparent responsiveness of said network system.

14. The method of claim 13, further comprising the steps of:

determining current utilization as the proportion of total bandwidth that is currently available and currently unavailable.

15. System for monitoring performance, capacity, and utilization of a network system including a test station, a network, and a target station, comprising:

means for performing test routines including test transactions across said network and probative tests of said network;

means responsive to data from said transactions and said tests for determining the state of the network, including as a function of network queue factor, network capacity and utilization;

means responsive to said state for deriving network contribution to application response time, and the capacity and utilization of said target station and its contribution to network system response time;

said means for performing test routines further including;

user input device at said test station for receiving from a user the name or network address of said target station and for customizing said test station as to number, size, and wait time between transmissions of test routines;

transmission means for sending said test transactions and probative tests to said target station, including discrete long pings, discrete short pings, burst of long pings, burst of short pings, first transaction samples that stress the processor of said target station, and second transaction samples that stress the storage retrieval function of said target station;

means for generating test results from said test transactions and probative tests including time of transmission from said test station and receipt back from said target station, number of bytes sent and received for each ping and transmission sample, and number of network sample frames dropped by the network; and

means for analyzing said test results, including calculating network latency, network response time bandwidth, network throughput bandwidth, network queue depth, and network utilization.

16. A system for monitoring performance, capacity, and utilization of a network system including a test station, a network, and a target station, comprising:

means for performing test routines including test transactions across said network and probative tests of said network;

means responsive to data from said transactions and said tests for determining the state of the network, including network capacity and utilization;

means responsive to said state for deriving network contribution to application response time, and the capacity and utilization of said target station and its contribution to network system response time;

said means for performing test routines further including;

user input device at said test station for receiving from a user the name or network address of said target station and for customizing said test station as to number, size, and wait time between transmissions of test routines;

transmission means for sending said test transactions and probative tests to said target station, including discrete long pings, discrete short pings, burst of long pings, burst of short pings, first transaction samples that stress the processor of said target station, and second transaction samples that stress the storage retrieval function of said target station;

means for generating test results from said test transactions and probative tests including time of transmission from said test station and receipt back from said target station, number of bytes sent and received for each ping and transmission sample, and number of network sample frames dropped by the network; and

means for analyzing said test results, including calculating network latency, network response time bandwidth, network throughput bandwidth, network queue depth, and network utilization;

means responsive to said network latency, network capacity, and said network utilization for calculating network response time for first and second test transactions of at least two message lengths;

means responsive to the response times for best and average network times calculated for said first and second test transactions for deriving average and best target system transaction time exclusive of network time;

means responsive to said best and average target system transaction time, for determining the average number of transactions in the queue of said target system; and

means for calculating transaction utilization as said number of transactions in said queue divided by one plus said number of transactions in said queue.

17. A method for monitoring performance, capacity, and utilization of a network system including a test station, a network, and a target station, comprising the steps of:

performing test routines including test transactions across said network and probative tests of said network;

responsive to data from said transactions and said tests, determining the state of the network, including network capacity and utilization as a function of network queue factor;

responsive to said state, deriving network contribution to application response time, and the capacity and utilization of said target station and its contribution to network system response time;

said step for performing test routines further including;

receiving from a user the name or network address of said target station and customizing said test station as to number, size, and wait time between transmissions of test routines;

sending said test transactions and probative tests to said target station, including discrete long pings, discrete short pings, burst of long pings, burst of short pings, first transaction samples that stress the processor of said target station, and second transaction samples that stress the storage retrieval function of said target station;

generating test results from said test transactions and probative tests including time of transmission from said test station and receipt back from said target station, the number of bytes sent and received for each ping and transmission sample, and the number of network sample frames dropped by the network; and

analyzing said test results, including calculating network latency, network response time bandwidth, network throughput bandwidth, the network queue depth, and network utilization.

18. A method for monitoring performance, capacity, and utilization of a network system including a test station, a network, and a target station, comprising the steps of:

performing test routines including test transactions across said network and probative tests of said network;

responsive to data from said transactions and said tests, determining the state of the network, including network capacity and utilization;

responsive to said state, deriving network contribution to application response time, and the capacity and utilization of said target station and its contribution to network system response time;

said step for performing test routines further including;

receiving from a user the name or network address of said target station and customizing said test station as to number, size, and wait time between transmissions of test routines;

sending said test transactions and probative tests to said target station, including discrete long pings, discrete short pings, burst of long pings, burst of short pings, first transaction samples that stress the processor of said target station, and second transaction samples that stress the storage retrieval function of said target station;

generating test results from said test transactions and probative tests including time of transmission from said test station and receipt back from said target station, the number of bytes sent and received for each ping and transmission sample, and the number of network sample frames dropped by the network;

analyzing said test results, including calculating network latency, network response time bandwidth, network throughput bandwidth, the network queue depth, and network utilization;

responsive to said network latency, network capacity, and said network utilization, calculating network response time for first and second test transactions of at least two message lengths;

responsive to the response times for best and average network times calculated for said first and second test transactions, deriving average and best target system transaction time exclusive of network time;

responsive to said best and average target system transaction time, determining the average number of transactions in the queue of said target system; and

calculating transaction utilization as said number of transactions in said queue divided by one plus said number of transactions in said queue.

19. A method for monitoring performance, capacity, and utilization of a network system including a test station, a network, and a target station, comprising the steps of:

performing test routines including test transactions across said network and probative tests of said network;

responsive to data from said transactions and said tests, determining the state of the network, including network capacity and utilization as a function of network queue factor;

responsive to said state, deriving network contribution to application response time, and the capacity and utilization of said target station and its contribution to network system response time; and

calculating average network message size and network utilization by;

calculating unadjusted standard deviation value Tq for a set of short ping tests and long ping tests; and

adjusting sigma Tq and queue depth to account for variance in service times as well as the multihop and multi server nature of the network.

20. A method for monitoring performance, capacity, and utilization of a network system including a test station, a network, and a target station, comprising the steps of:

performing test routines including test transactions across said network and probative tests of said network;

responsive to data from said transactions and said tests, determining the state of the network, including network capacity and utilization;

responsive to said state, deriving network contribution to application response time, and the capacity and utilization of said target station and its contribution to network system response time;

calculating average network message size and network utilization by;

calculating unadjusted standard deviation value Tq for a set of short ping tests and long ping tests; and

adjusting sigma Tq and queue depth to account for variance in service times as well as the multihop and multi server nature of the network;

calculating sigma Tq by adjusting said unadjusted Tq for the number of servers discovered in network testing;

further adjusting sigma Tq for network message size variance;

responsive to tests including isolated pings, determining network average one way queue delay;

responsive to tests including bursts of pings, determining the streaming bandwidth of said network;

responsive to said queue delay and said streaming bandwidth, deriving the network queue depth value; and

adjusting said network queue value for multiple hops so that the value for adjusted sigma Tq closely approaches the value for Tq.

21. Method for determining utilization of a network system, comprising the steps of:

transmitting and receiving back a plurality of short ping packets in isolation from each other with respect to a target system across a network;

transmitting and receiving back a plurality of long ping packets with respect said target system across said network;

recording transmission and receipt time stamps for said packets, and calculating the round trip time for each said packet;

determining the average round trip times and the best and the worst times for said short ping packets and for said long ping packets;

calculating apparent network bandwidth;

determining network streaming bandwidth;

determining network duplex factor, hop count, and multi-server factor; and

determining number of dropped short ping and long ping test packets.

22. Method for monitoring performance of a multi-server network system, including the steps of:

determining the average number of bits in network queues;

determining average network message size;

responsive to said average number of bits and said message size, calculating network utilization; and

measuring queue buildup as u^n/(1−

u^n), where u is utilization and n is the number of servers in said multi-server network.

23. The method of claim 22, said step for measuring queue buildup further comprises the steps of performing discrete echo test, throughput test, hop count measurement, throughput factor determination, duplex factor determination, and multi-server factor determination.

24. A method for monitoring performance, capacity, and utilization of a network system, comprising the steps of:

transmitting through said network system pings of different sizes to generate test data;

responsive to said test data, determining apparent responsiveness of said network system;

transmitting test packets among devices in or attached to said network to generate test results, including a plurality of tests selected from the set of tests including;

sending echo or discard packets (e.g., pings) of uniform length, isolated from one another by fixed intervals;

sending echo or discard packets of uniform length in a stream;

sending echo or discard packets of different lengths, isolated from one another by fixed intervals;

sending a file or equivalent batch transfer unidirectionally across the network, repeating with different packet sizes;

sending a file bidirectionally across the network, repeating with different packet sizes;

sending multiple files unidirectionally across the network; and

/or

sending multiple files bidirectionally across the network;

sending processor-bound transactions across the network to end systems;

sending I/O-bound transactions across the network to end systems; and

sending general transactions across the network to end systems.

25. The method of claim 24, further including the step of specifying a different transmission priority for each of at least two of said test packets.

26. The method of claim 25, further comprising the steps of:

first detecting whether prioritization has effect in the network;

determining the utilization of the network at different priority levels; and

then

deriving predictive results for current and future response time and window sizes for different types of service.

27. The method of claim 24, further comprising the steps of:

for each of a plurality of test transmissions, specifying a transmission priority that varies between different types of tests or between multiple iterations of one particular type of test;

applying network queue factor to the results of the said tests to mathematically evaluate the network under test with respect to capacity network response in transaction-type traffic, in screen image traffic, in batch traffic, in video, voice and other real-time traffic.