« PreviousContinue »
METHOD AND APPARATUS FOR NON-INTRUSIVE MEASUREMENT OF ROUND TRIP DELAY IN COMMUNICATIONS NETWORKS
BACKGROUND OF THE INVENTION
1. Technical Field
The present invention pertain to a method and apparatus io for measuring round-trip delay or travel time in a communications network while the communications network is in-service.
2. Discussion of the Prior Art
Numerous prior art communications networks send packets of information to target sites. Such networks are typically wide area networks (WAN) and include Frame Relay, ATM, SMDS, X.25 and ISDN. Long delay or travel times for packets tend to indicate an inefficient route, problems in the 2Q network, or other deficiencies which must be addressed to improve performance. Measurement of delay or travel time is conventionally accomplished by sending protocol specific packets to certain destinations capable of recognizing those packets, and then generating and transmitting a response 25 back to the originator. The round-trip delay or travel time is measured as the absolute time from generation of the protocol specific packet to receipt of the response.
Prior art measurement of delay or travel time through communications networks suffers from major disadvan- 30 tages. In particular, the measured delay or travel times include the processing delay introduced by recognition of the protocol specific packet and subsequent generation of the response. The protocol specific packets increase traffic on the network, thereby adding to the effective delay or travel 35 time in the system. Further, delay or travel time measurements may only be made when the available protocols and equipment are capable of supporting responses to certain protocol specific packets.
OBJECTS AND SUMMARY OF THE
Accordingly, it is an object of the present invention to provide a method and apparatus for measuring the round-trip 45 delay or travel time in a communications network without taking the communications network out of service, and excluding variable delays imposed by protocol processing at the end points.
It is another object of the present invention to provide a 50 method and apparatus for measuring the round-trip delay or travel time in a communications network without requiring measurement-specific data but instead processing the actual data stream that happens to be transmitted in the network, regardless of the protocol. 55
According to the present invention, probes are situated at monitor points of interest in a communications network. Typically, although not necessarily, the probes are located at actual target sites of the network. During normal in-service operation, identifiable data patterns are sent between sites, 60 traversing the monitor points. Each probe captures identifiable data patterns arriving at and departing its respective monitor point, and generates a time stamp indicating the time of the arrival or departure. The probe also generates a pattern identifier derived from the data in the identifiable 65 data pattern to uniquely identify that identifiable data pattern, and stores the time stamp and corresponding pattern
identifier in an internal buffer. Once the internal buffers contain a predetermined number (preferably user-specified) of identifiable data patterns, the pattern identifiers collected in the buffers of the two probes are matched by a processor to coordinate arrival and departure times of each identifiable data pattern. If fewer matches than a predetermined number of matches are found, the measurement of the round-trip delay or travel time fails. If the number of matches found equals or exceeds the predetermined number, the processor calculates an average round-trip delay or travel time based on the arrival and departure times of identifiable data patterns traveling in both directions between monitor points.
The above and still further objects, features and advantages of the present invention will become apparent upon consideration of the following detailed description of a specific embodiment thereof, particularly when taken in conjunction with accompanying drawings wherein like references in the various figures are utilized to designate like components.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 is a block diagram of a system for measuring the round-trip delay or travel time according to the present invention.
FIG. 2 is a software flow chart for controlling measurement of round-trip delay or travel time according to the present invention.
FIG. 3 is a functional block diagram of a probe employed in the system of FIG. 1.
FIG. 4 is a software flow chart illustrating the operation of the probe of FIG. 3.
FIG. 5 is a software flow chart illustrating the process for coordinating arrival and departure times in the system of FIG. 1.
FIG. 6 is a functional block diagram of the system of FIG. 1, further illustrating time stamp generation.
DESCRIPTION OF THE PREFERRED
Referring to the system illustrated in FIG. 1, two sites (A and B) are connected in a packetized data communications network by communication lines 10, 11. The preferred embodiment utilizes a packetized data communications network, however the present invention is applicable to any communications network where the data contains identifiable patterns and is not altered by the network. Each site is capable of both transmitting and receiving data packets conforming to whatever data protocol maybe employed in the network. Each communication line 10, 11 represents a respective transmission direction as indicated by the arrows. Two probes 12,13 are each connected to both communication lines 10, 11 at respective monitoring points Am, Bm to capture and process data packets being sent between sites A and B. The far sides of the respective probes 12 and 13 capture data packets arriving at the probe, while the near sides capture data packets departing from the respective probes 12 and 13. The terms "arriving" and "departing" are employed to imply transmission direction and are truly accurate only in the typical case where probes 12, 13 are situated at the actual sites A and B. However, the probes may be situated at any points in between to measure the roundtrip delay or travel time between those points. When probes 12, 13 have captured a user-specified amount of data, the processed data is then transmitted over secondary communication paths 14,15 to a console 16 which internally stores and analyzes the processed data to produce the round-trip
delay or travel time measurement. Typically, console 16 is a conventional personal computer or other general purpose computer and may be situated at any location having a communication capability with the probes.
The operation of the overall system is described with reference to FIGS. 1 and 2. A system operator requests console 16 to perform a round-trip delay or travel time measurement. The console configures and triggers probes 12, 13 to start capturing and processing data. Console 16 then polls probes 12, 13 to determine if a predetermined (typically user-specified) amount of data has been captured by each probe. If the predetermined amount of data has been captured by both probes, the processed data is transmitted via paths 14,15 and internally stored in console 16. Console 16 analyzes the received data to determine whether the data is capable of producing an accurate measurement as described below. If the received data is capable of producing an accurate measurement, console 16 calculates the roundtrip delay or travel time.
Referring to FIG. 3, probes 12, 13 are shown in block diagram form. Specifically, signals from communication lines 10, 11 are received at probe 12, for example. The signals are applied directly to respective line interface circuits 31, 32 that function to adapt the signals to standard digital logic levels used by respective packet receivers 33, 34. The packet receivers identify individual data packets within a data stream and store copies of the packets in a packet RAM 35. Packet RAM 35 is coupled to and shared by microprocessor 36 which processes the packet and generates corresponding time stamps and data packet identifiers 30 in the manner described below. In parallel to these operations, microprocessor 36 responds to requests from console 16.
Processing of the received data by microprocessor 36 is described in relation to FIGS. 3 and 4. Specifically, microprocessor 36 retrieves each data packet from packet RAM 35 and compares the transmission destination address for the packet with an address established as a parameter when console 16 configures the probes in response to a user request for a time delay measurement. If the transmission destination address of the retrieved packet matches the parameter address, the packet is determined to be part of the data traversing the monitoring points for which measurement is desired, and a timestamp and packet identifier are generated and stored in a table in a buffer internal to the microprocessor 36. Packets with transmission destination addresses not matching the parameter address are ignored. The format of tables stored in the internal buffer of microprocessor 36 is illustrated in Table I. The tables comprise two columns representing the packet identifiers and timestamps.
Each individual table in Table I contains packets received exclusively on one of the near or far lines of the probe (FIG. 1). In the preferred embodiment, each probe contains a single table, typically holding two hundred entries, where the most significant bit of the packet identifier indicates the , side on which the packet is received as described below. The process continues until console 16 requests the data of the buffers in the probe when the total number of datum exceeds a predetermined number. The process may also terminate when the buffer is full and console 16 does not request data.
Microprocessor 36 generates a time stamp by maintaining and sampling a 32 bit free-running counter clocked by 1 KHz pulses derived from the microprocessor 24.704 MHz crystal oscillator. The counter has a one millisecond resolution and has a sufficiently large capacity that rollover infrequently occurs. The counters of each probe need not be synchronized, but they must be clocked at substantially identical rates to assure measurement accuracy.
In the preferred embodiment, the packet identifiers generated by microprocessor 36 (FIG. 3) comprise packet signatures. The signature for a packet is created by utilizing the data in that packet. Specifically, a user-specified number of bytes of data at the beginning of the packet are omitted while the remainder of the packet is treated as an array of 32-bit unsigned integers (i.e., the most significant bit is treated as part of the integer and does not designate the sign of the integer). The user-specified number of bytes are set as a parameter during configuration of the probes by console 16 (FIG. 3). The omitted bytes at the beginning of the packet are omitted in order to avoid inconsistent signatures due to modification of packet data by transmission systems during transmission. The signature is obtained by adding the 32-bit unsigned integers together, ignoring overflow, with any remaining bytes left over in the packet. The packet length may be added to this sum to generate unique signatures if the packets contain mostly zeros. The packet length is treated as an unsigned 16-bit integer. The most significant bit of the signature is set to one to indicate the packet was retrieved from the near side of the probe, and set to zero to indicate retrieval from the far side.
The unique identifier for the packet is generated in order to later coordinate packet arrival and departure times. When console 16 requests data, the data in the buffer of the probe is released to ethernet interface 37 to make the data compatible for transfer over ethernet bus 38 to console 16 for processing.
The above-described components 31-37 of probes 12,13 are all conventional and commercially available. The preferred embodiment utilizes the following integrated circuits with corresponding circuitry: Level One LXT901PC as