Background of the Invention
1. Field of the Invention
The present invention relates to a method and apparatus for monitoring data transmission through a communications network while the communications network is in service. More particularly, the present invention relates to a network monitoring system having endpoint probes that transmit and receive message data over a packetized switching network to exchange information allowing the probes to measure network performance metrics, such as network availability, data delivery ratio and round trip delay in the communications network.
2. Description of the Related Art
Both from an end user and service provider standpoint, there is an increasing need to accurately measure operational performance of data communications networks. Communications networks, especially packetized data networks, are currently utilized in various applications for transmission and reception of data between parties at different locations. A typical data transmission system includes a plurality of end user sites and a data packet switching network, which resides between the sites to facilitate communications. Each site is connected to the switching network via an access channel (i.e., a channel connecting a site to a communications system), wherein transmission circuits, preferably virtual circuits, establish paths between the sites through the access channel and the switching network.
Packetized data networks typically format data into packets for transmission from one site to another. In particular, the data is partitioned into separate packets at a transmission site, wherein the packets usually include headers containing information relating to packet data and routing. The packets are transmitted to a destination site in accordance with any of several conventional data transmission protocols known in the art (e.g., Asynchronous Transfer Mode (ATM), Frame Relay, High Level Data Link Control (HDLC), X.25, IP tunneling, etc.), by which the transmitted data is restored from the packets received at the destination site.
Packetized data communications are especially appealing for common carrier or time-shared switching systems, since a packet transmission path or circuit is unavailable only during the time when a packet utilizes the circuit for transmission to the destination site, thereby permitting other users to utilize that same circuit when the circuit becomes available (i.e., during intervening periods between packet transmissions). The access channel and each individual transmission circuit typically have a maximum data carrying capacity or bandwidth that is shared among the various users of the network. The access channel utilization is typically measured as an aggregate of the individual circuit utilizations and has a fixed bandwidth, while the individual circuits may be utilized by several users wherein each user may utilize an allocated portion of the circuit.
Typically, when a party needs to send and receive data over distances, the party (end user) enters into a service contract with a service provider to provide access to a data communications network. Depending on an individual end user's needs, the service contract may include provisions that guarantee certain minimum performance requirements that the service provider must meet. For example, if the end user expects to send and receive a certain amount of data on a regular basis, the end user may want the service provider to guarantee that a certain minimum bandwidth will be available to the end user at all times. Certain end user applications are sensitive to transmission delays and/or the loss of data within the network (i.e., failure to successfully deliver data packet(s) to their destination). Specifically, while loss of data packets can generally be detected by end users (via information provided in the data transmission protocol), and lost packets can be retransmitted, certain applications cannot function when the percentage of lost data exceeds a given level. Thus, the end user may want the service provider to guarantee that the average or minimum ratio of data units delivered by the network to data units offered to the network at the far-end is above a certain percentage and/or that the average or maximum transmission delays will not exceed a certain duration.
From a service provider's perspective, it would be competitively advantageous to be able to demonstrate to potential and existing end users that the service provider is capable of meeting and does meet such network performance metrics. Thus, the capability to provide analysis of network system performance at the service level, i.e., service level analysis (SLA), particularly in the context of network systems that share bandwidth between sites, would be advantageous from both an end user and service provider standpoint.
Various systems have been proposed which provide some measure of network system performance. Specifically, a number of techniques for measuring round trip delay (RTD) of data transmitted between two sites is known. For example, U.S. Pat. No. 5,521,907 to Ennis, Jr. et al., the disclosure of which is incorporated herein by reference in its entirety, discloses a system for passively measuring the round trip delay of data messages sent between two sites. More specifically, a console triggers probes at two sites to store data packets being sent between the two sites. The probes generate unique packet signatures based on the data in the packets, and time stamp the signatures. By matching signatures from the two probes and comparing the corresponding timestamp values, the console can determine the round trip delay between the sites. This technique requires the storage, transmission and processing of a significant amount of data, particularly if implemented to periodically monitor all virtual circuits existing between a set of sites. That is, the passive probes cannot individually determine round trip delay, and each probe must store and transmit a substantial amount of data to the console which is required to correlate signature and timestamp data from different sites.
U.S. Pat. No. 5,450,394 to Gruber et al., the disclosure of which is incorporated herein by reference in its entirety, discloses a technique for determining round trip delay in which measurement cells containing timestamp information are sent between two nodes. A first node transmits a measurement cell with a first time stamp to a second node, and the second node replies with a measurement cell containing additional time stamp information which can be used by the first node to determine the round trip delay. Because the technique relies, in part, on timestamps already present in PM OAM (performance management operations, administration and maintenance) ATM cells, the technique is specific to the ATM protocol and cannot readily be adapted to other data protocols or be expanded to monitor other service level performance metrics. Further, the technique does not allow both nodes to measure the round trip delay of the same sequence of cells (i.e., either only one of the two nodes measures round trip delay or the two node measure delays of different transmitted cell sequences).
Further, while it is possible for individual switches in existing network systems to indicate how many packets of data have been dropped by the switch, there are no known systems capable of measuring a rate of successful (or unsuccessful) data delivery on a service level, e.g., over a particular virtual circuit or to a particular end user.
The problem of providing service level analysis of network performance is complicated by the fact that many switching networks comprise interworked systems using plural, different data transmission protocols (e.g., an ATM switching network interworked with a Frame Relay switching network), thereby forming a so-called "interworked" network. Such interworked networks are becoming more common, and present an additional challenge to designing a service level analysis tool that employs a standard message structure and messaging protocol useful for communicating between any two sites. Existing systems relying on inter-site or inter-probe messages to assess system performance are generally incapable of operating across interworked networks.
Accordingly, there remains a need for a system capable of providing service level analysis (SLA) of communications network performance, especially packetized, interworked data networks, to provide end users and service providers information relating to performance metrics, such as round trip delay, data delivery ratio, and other metrics, such as the percentage of time the network is available.
Summary of the Invention
It is an object of the present invention to provide service level analysis of network performance metrics of interest to data communications network end users and service providers.
It is a further object of the present invention to assess whether communications network transmission delays remain within acceptable limits.
It is yet a further object of the present invention to monitor the round trip delay of data transmitted over a transmission circuit (e.g., a virtual circuit (VC) or a permanent virtual circuit (PVC)) connecting endpoint sites.
It is a still further object of the present invention to assess whether a data communications network delivers to a destination, an acceptable percentage of data units offered to the network that are bound for the destination, particularly in a shared bandwidth environment.
It is another object of the present invention to monitor a rate of successful (or unsuccessful) data delivery of data transmitted over a transmission circuit connecting endpoint sites.
It is yet another object of the present invention to actively monitor network performance metrics, such as round trip delay and data delivery ratio, by transmitting over a transmission circuit connecting endpoint sites, messages containing data transmission monitoring information.
It is still another object of the present invention to provide service level analysis of network performance metrics in an interworked network environment.
A further object of the present invention is to provide a tool capable of performing service level analysis with any conventional data transmission protocol.
Yet a further object of the present invention is to transmit service level analysis (SLA) messages over transmission circuits connecting user endpoints that can be encapsulated in a single, standard data unit of any conventional data transmission protocol.
Still a further object of the present invention is to provide service level analysis of network performance metrics by transmitting SLA messages over circuits connecting endpoint sites without requiring synchronization between local site/probe clocks.
Another object of the present invention is to monitor network availability of transmission circuits connecting endpoint sites.
The aforesaid objects are achieved individually and in combination, and it is not intended that the present invention be construed as requiring two or more of the objects to be combined unless expressly required by the claims attached hereto.
The system of the present invention provides a service level analysis (SLA) capability for communications networks, especially any packetized, circuit-oriented data networks, wherein packets (or units) of data are transmitted to a destination site in accordance with one or a combination of data transmission protocols, including, but not limited to: Frame Relay; ATM and X.25. According to the present invention, a data transmission system includes endpoint probes connected between corresponding endpoint user sites and a data switching network. Each probe is connected to the switching network via an access channel wherein transmission circuits, preferably virtual circuits (e.g., a permanent virtual circuit (PVC), a switched virtual circuit (SVC) or a tunnel through an IP network), establish paths between the sites/probes through the access channel and switching network. A virtual circuit is basically a path established in a packet switching network to transfer data to a specific destination or site.
Each probe captures and retransmits data traveling between its site and other sites over respective transmission circuits (such as virtual circuits (VCs)) of the switching network, and can thereby insert service level analysis (SLA) messages into the data traffic in order to actively generate and communicate network performance information to other probes.
Each probe periodically stores data collected for each VC associated with the site, which data is used to monitor certain network performance metrics on a per-VC basis. Specifically, each probe performs service level analysis, including monitoring of one or more of the following system performance metrics: round-trip delay, data delivery ratio and availability. These performance metrics are measured for each virtual circuit connecting endpoint sites through the network, although these metrics are affected by the access line and access channel.
During each SLA measurement cycle, whose duration can be set at a desired predetermined value, such as every fifteen minutes, a probe exchanges a sequence of SLA messages over each VC connecting the probe to other probes. The sequence of messages contains timestamp and counter data that allows the probe to determine the round trip delay of the SLA messages as well as the number of data units offered to the network (bound for the probe's site) and the number of data units actually delivered to the site for each VC connected to the site. The probes are not required to be time synchronized in order to exchange these SLA messages.
In accordance with another aspect of the present invention, the SLA message contents and protocol are designed to allow the SLA messages to be encapsulated in a single, standard data unit of any conventional data transmission protocol, e.g., an ATM cell or a Frame Relay frame. Consequently, the messaging system can be used with any data transmission protocol and in interworked networks without modification of the message data payload.
The SLA measurements collected by the probes can be transmitted to a console in communication with the probes for processing, display, and archiving. The SLA measurements provided by the probes of the present invention allows end users and service providers to know whether the network performance is meeting requirements set forth in a customer service agreement.
The above and still further objects, features and advantages of the present invention will become apparent upon consideration of the following detailed description of specific embodiments thereof, particularly when taken in conjunction with the accompanying drawings wherein like reference numerals in the various figures are utilized to designate like components.
Brief Description of the Drawings
FIG. 1 is a functional block diagram of a data transmission system having service level analysis probes disposed between endpoint sites and a switching network in accordance with an exemplary embodiment of the present invention.
FIG. 2 is a diagram illustrating the time sequence and contents of a sequence of inter-probe messages sent between two probes of the present invention.
FIGS. 3A and 3B illustrate a functional flow chart corresponding to the diagram of FIG. 2, indicating operations performed by the probes of the present invention in order to capture and store measurements supporting service level analysis of network performance metrics.
FIG. 4 is a diagram illustrating an inter-probe message structure used in accordance with an exemplary embodiment of the present invention to transmit measurements supporting service level analysis of network performance metrics.
Detailed Description of the Preferred Embodiments
A system for monitoring performance and providing service level analysis (SLA) capability for data communications networks is illustrated in FIG. 1. Specifically, an exemplary data transmission system 10 includes two sites (A and B) and a packet switching network 12 to facilitate communications between the sites. Site A is connected to network 12 via a probe A, while site B is connected to network 12 via another probe B. Site A is connected to probe A by communication lines 20 and 21, probe A is connected to network 12 by communication lines 22 and 23, network 12 is connected to probe B by communication lines 24 and 25, and probe B is connected to site B by communication lines 26 and 27. The data transmission system 10 typically includes conventional telecommunications line types, such as T3, OC-3, North American T1 (1.544 Mbits/second), CCITT (variable rate), 56K or 64K North American Digital Dataphone Service (DDS), and a variety of data communications connections, such as V.35, RS-449, EIA 530, X.21 and RS-232. Sites A and B are each capable of transmitting and receiving data packets in various protocols utilized by communication lines 20, 21, 26 and 27, such as Asynchronous Transfer Mode (ATM), Frame Relay, High Level Data Link Control (HDLC) and X.25. Each line 20-27 represents a respective transmission direction as indicated by the arrows. For example, the arrows on communication lines 20 and 27 represent transmissions from sites A and B to probes A and B, respectively, while the arrows on communication lines 21 and 26 represent transmissions to sites A and B from probes A and B, respectively. Similarly, the arrows on communication lines 22 and 25 represent transmissions from probes A and B to switching network 12, respectively, while the arrows on communication lines 23 and 24 represent transmissions to probes A and B from switching network 12, respectively.
Generally, site A and site B utilize switching network 12 to communicate with each other, wherein each site is connected to switching network 12 via an access channel having transmission circuits, preferably virtual circuits, that establish paths between the sites through the access channel and switching network. The access channel refers to the lines utilized by each site to communicate with the switching network (i.e., communication lines 20-27), while a virtual circuit is basically a path established through a packetized data switching network that transfers data to a specific endpoint or site.
As used herein, the term "packet" (e.g., as used in "packetized switching network") does not imply any particular transmission protocol and can refer to units or segments of data in a system using, for example, any one or combination of the above-listed data transmission protocols (or other protocols). However, since the term "packet" is often associated with only certain data transmission protocols, to avoid any suggestion that the system of the present invention is limited to any particular data transmission protocols, the term "protocol data unit" (PDU) will be used herein to refer the unit of data being transmitted from sites A and B through switching network 12. Thus, for example, a PDU can be carried on a frame in the Frame Relay protocol, a related set of cells in the ATM protocol, or a packet in the IP protocol.
As shown in FIG. 1, probes A and B are respectively disposed between switching network 12 and sites A and B. Probes A and B can be located at sites A and B or at any point between switching network 12 and sites A and B. Probes A and B allow PDUs being sent between sites A and B via switching network 12 to pass through the probe and also insert inter-probe message PDUs into the data traffic. As used herein, the term "to the switching network" refers to the direction of data traveling to the switching network (e.g., data traveling on communication lines 20, 22, 25 and 27), while the term "from the switching network" refers to the direction of data traveling from the switching network (e.g., data traveling on communication lines 21, 23, 24 and 26). The terms "arriving", "to switching network", "departing", and "from switching network" are all relative and are employed to imply transmission direction.
For illustrative purposes, only two sites (A and B) are shown in FIG. 1. However, it will be understood that the data communication system can include numerous sites, wherein each site is generally connected to multiple other sites over corresponding transmission circuits, such as virtual circuits (VCs).
In accordance with an exemplary embodiment of the present invention, probes A and B actively perform service level analysis (SLA) by exchanging a sequence of inter-probe SLA messages that are inserted into the data traffic traveling over switching network 12 between sites, and that contain information allowing the probes to monitor one or more of the following system performance metrics on a per-virtual-circuit basis for each virtual circuit connecting each site to other sites: round trip delay, data delivery ratio and network availability. Round-trip delay (RTD) can be defined as the duration of time required for a PDU to go from a first end of a virtual circuit (VC) to the second end plus the time required for return PDU to go from the second end of the VC to the first end. This round trip delay does not include the far-end turn around time (i.e., the duration between the receive time of the first PDU at the second end and the transmit time of the return PDU from the second end). Optionally, components of the delay that are not attributable to the switching network, such as those related to the access line rate (i.e. serialization delays), can be excluded from the computed RTD. Data Delivery Ratio (DDR) is the ratio of the number of PDUs delivered on a transmission circuit by the network, to the number of PDUs offered to the network on that transmission circuit. The DDR can be computed from the aggregate of the PDUs offered and delivered in both directions of the circuit or as separate, one-way ratios, as described in greater detail below. Availability is defined as the percent of the time that the network (or virtual circuit) is capable of accepting and delivering user PDUs within the total time of a measurement period. Unlike RTD and DDR, Availability can be determined passively, without inserting messages into the data traffic, as explained below.
The SLA measurements performed by the probes to monitor these network performance metrics apply specifically to the virtual circuits through the switching network, although they are affected by the access line and access channel. The measurements that are accumulated on the probes of the present invention support service level analysis. Each measurement is preferably accumulated on a continuous basis with a minimum of one historical sample per fixed historical time interval, e.g., once every fifteen minutes. As will become evident, the measurements are collected in a manner that scales to very large networks without performance degradation.
In order to acquire the service level analysis (SLA) measurements required to determine round trip delay (RTD) and data delivery ratio (DDR), endpoint probes A and B insert data into the user data stream being transmitted through the network over the managed transmission circuit connecting sites A and B, which circuit can be a virtual circuit (VC). Probes A and B also insert similar data over the other VCs connecting sites A and B to other sites.
More specifically, each site's probe separately maintains a periodic historical measurement cycle (which can be, for example, 15 minutes long) which is not required to be synchronized with the periodic measurement cycles of other probes. During each periodic cycle, each probe acquires one set of SLA measurements for each VC. At an activation time within the cycle (which is an arbitrary, predetermined time after the beginning of the measurement cycle, e.g., the mid-point of each measurement cycle), each probe initiates an inter-probe message sequence on each VC that has not already had an inter-probe message sequence initiated by a far-end probe. That is, because the periodic cycles of the endpoint probes in the communications system are not synchronized, when the activation time within a cycle occurs for a particular probe (probe A), certain far-end probes connected to probe A over certain VCs may have already reached the activation times within their measurement cycles, and therefore would have initiated a message sequence for these VCs. Other far-end probes connected to probe A over other VCs may not have reached the activation times within their cycles at the activation time of probe A's cycle, and therefore would not have initiated a message sequence for these other VCs. For illustrative purposes, it is assumed in the exemplary embodiment that probe A reaches the activation time within its periodic measurement cycle prior to probe B reaching the activation time within its periodic measurement cycle, and therefore probe A initiates the inter-probe messaging sequence with probe B over the VC connecting probes A and B. However, it will be understood that the hereafter-described message sequence is initiated for every VC by the one of the VC's two endpoint probes that first reaches the activation time within its periodic historical measurement cycle.
Each probe acquires certain information for each VC during each measurement cycle in order to support SLA. Specifically, in order for each probe to be able to determine the round trip delay (RTD) for a VC, the probe must know the transmit and receive times of an SLA message PDU transmitted in one direction over the VC and the transmit and receive times of a reply SLA message PDU transmitted in the opposite direction over the VC.
In order to determine the data delivery ratio (DDR), for each monitored VC, each probe maintains a counter that accumulates the count of the total number of PDUs delivered ("delivered PDUs") by the network over the VC (e.g., probe A maintains a count of the number of PDUs delivered by the network from probe B and vise versa). Additionally, each probe maintains a counter, per VC, that accumulates the total number of PDUs offered by the probe.
To support SLA, probe A acquires and stores in each measurement cycle a count of the PDUs bound for site A that were offered to the network by probe B (not initially known by probe A, but known by probe B), and a count of the PDUs originating from site B that were delivered by the network to probe A (known by probe A from its counters). Similarly, probe B acquires and stores in each measurement cycle a count of the PDUs bound for site B that were offered to the network by probe A (not initially known by probe B, but known by probe A), and a count of the PDUs originating from site A that were delivered by the network to probe B (known by probe B from its counters). Probes A and B acquire these measurements for all other VCs as well.
Referring to FIGS. 2 and 3, acquisition of the aforementioned SLA measurements by endpoint probes A and B is described. When probe A reaches the activation time of a periodic measurement cycle, probe A sends an "initial request" message M.sub.1 to probe B (see FIG. 2, FIG. 3A, blocks 40-44). The actual transmit time T.sub.1 of message M.sub.1 is not known by probe A until the message M.sub.1 is sent; accordingly, message M.sub.1 does not contain timestamp information indicating the time of its transmission.
Once the transmit time T.sub.1 of message M.sub.1 is known, probe A records a count (OFF.sub.A (i)) of the PDUs (bound for probe B) that probe A has offered to the network up to time T.sub.1. The index i denotes the present periodic measurement cycle. The PDU count OFF.sub.A (i) can be a "delta" count, meaning that the count is the number of PDUs offered to the network from the time the last initial request message was sent (at the activation time within the previous (i-1) periodic historical measurement cycle) up to time T.sub.1. However, for reasons that will become evident, the PDU count OFF.sub.A (i) and the other PDU counts recorded by the probes are preferably "raw" counts, meaning that the counts are running totals counted from the time of system boot up or the last counter roll-over. The use of raw counts requires the probes to maintain a record of the raw counts recorded during the message sequence in the previous measurement cycle (i-1) in order to compute the delta PDU counts that pertain to the present data collection interval (which runs from the present activation time to the previous activation time).
As explained above, the count (OFF.sub.A (i)) is needed by probe B to support SLA; however, since OFF.sub.A (i) is not known until after the transmission time T.sub.1 of message M.sub.1 is known, OFF.sub.A (i) (like T.sub.1) cannot be sent to probe B in message M.sub.1. Time T.sub.1 is measured relative to a local clock of probe A, which clock need not be synchronized with a corresponding local clock of probe B.
As shown in FIG. 2, message M.sub.1 travels over the network VC and is received by probe B at time T.sub.2 (as measured relative to a local clock of probe B). In response to reception of message M.sub.1, probe B records a count (DEL.sub.B (i)) of the PDUs sent by probe A that have been delivered by the network to probe B up to time T.sub.2 (see FIG. 3A, block 46). Notably, because this PDU count DEL.sub.B (i) and the PDU count OFF.sub.A (i) measured by probe A are defined by the transmit and receive times of the same message (M.sub.1), these counts relate to correlated time periods at probes A and B; thus, these counts can be directly compared (once converted to delta counts) to obtain a meaningful measure of one-way (A to B) data delivery ratio over the VC during a given data collection period.
In response to reception of initial request message M.sub.1, probe B sends a "reply.sub.B " message M.sub.2 to probe A (see FIG. 2, FIG. 3A, blocks 48) at time T.sub.3. The actual transmit time T.sub.3 of message M.sub.2 is not known by probe B until the message M.sub.2 is sent; accordingly, message M.sub.2 does not contain timestamp information indicating the time of its transmission. However, because the receive time T.sub.2 of initial request message M.sub.1 is known by probe B, a timestamp indicating the value of time T.sub.2 is sent by probe B to probe A in reply.sub.B message M.sub.2.
Once the transmit time T.sub.3 of message M.sub.2 is known, probe B records a count (OFF.sub.B (i)) of the PDUs (bound for probe A) that probe B has offered to the network up to time T.sub.3 (see FIG. 3A, block 48). As explained above, this count (OFF.sub.B (i)) is needed by probe A to support SLA; however, since OFF.sub.B (i) is not known until after the transmission time T.sub.3 of message M.sub.2 is known, OFF.sub.B (i) (like T.sub.3) cannot be sent to probe A in message M.sub.2. Like time T.sub.2, time T.sub.3 is measured relative to the local clock of probe B.
As shown in FIG. 2, message M.sub.2 travels over the network VC and is received by probe A at time T.sub.4 (as measured relative to probe A's local clock). In response reception of message M.sub.2, probe A records a PDU count (DEL.sub.A (i)) of the PDUs sent by probe B that have been delivered by the network to probe A up to time T.sub.4 (see FIG. 3A, block 50). Because this PDU count DEL.sub.A (i) and the PDU count OFF.sub.B (i) measured by probe B are defined by the transmit and receive times of the same message (M.sub.2), they relate to correlated time periods at probes A and B; thus, these counts can be directly compared (once converted to delta counts) to obtain a meaningful measure of one-way (B to A) data delivery ratio over the VC over a given data collection period.
As shown in FIG. 2, as this point in time (after T.sub.4), probe A knows the values of times T.sub.1, T.sub.2, T.sub.4, and PDU counts OFF.sub.A (i) and DEL.sub.A (i). To provide probe B with the aforementioned information necessary to support SLA, in response to reception of reply.sub.B message M.sub.2, probe A sends a "reply.sub.A " message M.sub.3 to probe B (see FIG. 2, FIG. 3A, block 52) at time T.sub.5 containing timestamps indicating the values of times T.sub.1 and T.sub.4 as well as the value of count OFF.sub.A (i). ReplY.sub.A message M.sub.3 travels over the network VC and is received by probe B at time T.sub.6 (FIG. 3A, block 54). At this point, probe B has acquired all of the measurements required to support SLA, i.e., the timestamps indicating the values of times T.sub.1 through T.sub.4 and PDU counts OFF.sub.A (i) and DEL.sub.B (i). Accordingly, probe B can compute the round trip delay (RTD) as the difference between the overall round trip time (T.sub.4 -T.sub.1) less the far-end turn around time (T.sub.3 -T.sub.2):
Note that this RTD calculation is valid despite the fact that probe A's reference clock is not synchronized with probe B's reference clock due to the subtraction of T.sub.1 from T.sub.4 and T.sub.2 from T.sub.3.
Further probe B can compute, from the raw PDU counts, the delta PDU counts for the present measurement cycle i (i.e., the counts of the PDUs accumulated in the collection data period ending at the time of the message exchange) as:
where OFF.sub.A (i-1) is the recorded (raw) count (sent to probe B in the last measurement cycle) of the number of PDUs bound for probe B that were offered by probe A to the network up to the transmit time of the initial request message M.sub.1 sent by probe A in the previous measurement cycle (i-1), and DEL.sub.B (i-1) is the recorded (raw) count of the number of PDUs sent by probe A that were delivered by the network to probe B up to the receive time of the initial request message M.sub.1 received by probe B in the previous measurement cycle (i-1).
As previously mentioned, PDU counts .DELTA.OFF.sub.A (i) and .DELTA.DEL.sub.B (i) can be directly compared to obtain a measure of the one-way VC data delivery ratio. For example, the one-way DDR can be computed as:
The data delivery ratio can also be computed as a two-way performance metric. In this case, probe B simply stores counts .DELTA.OFF.sub.A (i) and .DELTA.DEL.sub.B (i) for later processing by a console 16 (FIG. 1), which also receives counts .DELTA.OFF.sub.B (i) and .DELTA.DEL.sub.A (i) from probe A. Accordingly, at the end of each of probe B's periodic historical measurement cycle, the computed RTD and PDU delta counts .DELTA.OFF.sub.A (i) and .DELTA.DEL.sub.B (i) (and/or the one-way DDR) are stored by probe B for later transmission to console 16 (FIG. 3B, block 62).
To provide probe A with the aforementioned information necessary to support SLA, in response to reception of the reply.sub.A message M.sub.3, probe B sends a "final reply" message M.sub.4 to probe B (see FIG. 2, FIG. 3B, block 56) at time T.sub.7 containing a timestamp indicating the value of time T.sub.3 as well as the value of PDU count OFF.sub.B (i). Final reply message M.sub.4 travels over the network VC and is received by probe A at time T.sub.8 (FIG. 3B, block 58). At this point, probe A has acquired all of the measurements required to support SLA, i.e., the timestamps indicating the values of times T.sub.1 through T.sub.4 and PDU counts OFF.sub.B (i) and DEL.sub.A (i). Accordingly, probe A can compute the round trip delay (RTD) using equation (1). Note that probes A and B both measure the round trip delay of the same set of messages, i.e., messages M.sub.1 and M.sub.2.
Further probe A can compute the delta PDU counts for the present measurement cycle i as:
where OFF.sub.B (i-1) is the recorded (raw) count (sent to probe A in the last measurement cycle) of the number of PDUs bound for probe A that were offered by probe B to the network up to the transmit time of the reply.sub.B message M.sub.2 sent by probe B in the previous measurement cycle (i-1), and DEL.sub.A (i-1) is the recorded (raw) count of the number of PDUs sent by probe B that were delivered by the network to probe A up to the receive time of the reply.sub.B message M.sub.2 received by probe A in the previous measurement cycle (i-1).
PDU counts .DELTA.OFF.sub.B (i) and .DELTA.DEL.sub.A (i) can be directly compared to obtain a measure of the one-way VC data delivery ratio. For example, the one-way DDR can be computed as:
Again, the data delivery ratio can also be computed as a two-way performance metric. In this case, probe A simply stores counts .DELTA.OFF.sub.B (i) and .DELTA.DEL.sub.A (i) for later processing by a console, which also receives counts .DELTA.OFF.sub.A (i) and .DELTA.DEL.sub.B (i) from probe B. At the end of each of probe A's periodic historical measurement cycles, the computed RTD and PDU counts .DELTA.OFF.sub.B (i) and .DELTA.DEL.sub.A (i) (and/or the one-way DDR) are stored by probe A for later transmission to console 16 (FIG. 3B, block 60).
One advantage of transmitting raw PDU counts rather than delta PDU counts in the SLA message is that the loss of an SLA message PDU does not result in the loss of PDU count data. If a probe does not receive PDU count information within a measurement cycle due to a lost message PDU (message M.sub.3 or M.sub.4), the probe can simply wait until the next measurement cycle and compute the delta PDU counts for two consecutive data collection time intervals rather than the normal single data collection time interval.
Another advantage of maintaining raw PDU counts rather than delta PDU counts is that message contention situations are more easily managed. Specifically, it is possible that two probes initiate message sequences with each other substantially simultaneously (e.g., probe A transmits an initial request message M.sub.1 and probe B subsequently transmits an initial request message M.sub.1 prior to reception of probe A's message). Under these circumstances, when raw counts are used, the delta counts can be computed simply by subtracting the last-received PDU count values from the raw counts from the previous measurement cycle.
As an enhancement to the data delivery ratio measurement described above, the data delivery ratio can be determined for plural throughput ranges for certain data transmission protocols. For example, additional counters can be used to separately accumulate counts of PDUs where the throughput is below a committed data rate, and PDUs where the throughput is between the committed and burst date rate. This allows reports of DDR for each throughput range. The committed or peak rate of data flow can be specified differently for different transmission protocols. For example, for ATM, the peak rate of data flow is the Peak Cell Rate (PCR), and for Frame Relay the peak rate of data flow is, Bc+Be (where Bc is the number of bits the service provider is committed to deliver during a time interval, and Be is the number of bits in excess of Bc that the service provider might deliver during the time interval). Thus, for example, for the Frame Relay protocol, the count of non-burst offered frames (PDUs) can be expanded to add other two other counters: frames in seconds where the throughput is below Bc; and frames in seconds where the throughput is between Bc and Bc+Be.
Optionally, to minimize network impact, inter-probe messaging can be enabled/disabled on a per-VC basis. This allows the user to disable the message generation on VCs that do not have probes on each end.
In addition to measuring round trip delay (RTD) and data delivery ratio (DDR), the probes of the present invention can also calculate network availability. Unlike RTD and DDR, Availability is measured passively, i.e., without the assistance of SLA messages injected into the data traffic. Instead determination of Availability relies on analyzing information provided by the network operating under a particular protocol.
When the probe is operating under the Frame Relay data transmission protocol, the availability of a circuit can determined from the local management interface LMI activity between the endpoint user equipment and the network switches.
A circuit is deemed to be capable of accepting and delivering user frames whenever one of the following conditions is present on both of its ends: a local management interface (LMI) STATUS message is generated by the network in response to each STATUS ENQUIRY message, and the LMI STATUS message indicates that the circuit is valid and active; or No LMI STATUS ENQUIRY message has ever been seen by the probe (in other words, no LMI is in use on the link).
More specifically, at any one time, a circuit end-point can be in any one of the following states:
1. Link Down--user side (LMI requests are not being generated)
2. Link Down--network side (no LMI response to user generated requests)
3. Link Up, Data Link Connection Identifier (DLCI) invalid
4. Link Up, DLCI inactive
5. Link Up, DLCI active
From the service provider's standpoint, the circuit is considered available if both ends of the VC are in state 1 or 5. To facilitate this determination, the probe maintains a counter for each circuit, that accumulates a count of the number of seconds that that end of the circuit was ever not in either of these states. If a probe is assuming the role of the user's equipment when the user's equipment fails to participate, state 1 cannot occur because the probe itself guarantees that LMI requests are generated. Note that, because there is no required time synchronization between probes, the console cannot accurately "line up" periodic historical samples from both ends of the VC, so the console generally must pick one end of each circuit on which to base the availability (e.g., probe A in FIG. 1). Altematively, the console can query the probes at both ends of a circuit and use the weighted average of the two available times.
In the case of many Frame Relay switching networks, circuit status is transferred through the network so that, if the far-end site is in state 1, the near-end DLCI is marked inactive. If looking only at the near-end DLCI's statistics, the SLA report will show that the VC was down even though it was caused by far-end user equipment failure. If the probes act in the role of the user equipment, when it fails to LMI poll, this can be avoided.
In the case of ATM networks, availability cannot be derived from LMI messages. In some cases, LMI does not exist on an ATM network. OAM cell flows and cell delineation are the two factors that can be used to determine if a VC is active or not, and therefore available.
The first factor in availability is cell delineation. If the network is out of cell delineation, all VCs on that network are unavailable. The second factor in deciding availability is AIS or RDI cell flow. These OAM cells are sent into the access line on a VC to inform the network that service on that VC is not available. Any second where an AIS or RDI cell is received is considered an unavailable second.
Availability is measured on a per-circuit basis and stored in the probe as a count of the number of seconds in which the circuit was not available for any part of that second.
From time to time (e.g., a time period equal to or greater than periodic measurement cycle, up to several days), probes A and B download the aforementioned historical data samples taken at periodic intervals to a console (FIG. 1). For example, the probes collect the SLA parameters in the standard periodic measurement intervals with a two day history. Data can be periodically collected and stored in a database in accordance with a data collection process running on a database server.
The format and content of the measurements provided to the console for each periodic measurement interval is shown in Table 1.
Referring to FIG. 1, console 16 is in communication with probe A via a communication line 28 (e.g., a local area network (LAN)) and with probe B via a communication line 29, and retrieves the information collected by probes A and B to process the information for display and/or recordation. While shown for illustrative purposes as separate communication lines, it will be understood that the communication lines connecting probes A and B can be virtual circuits over the switching network or any other suitable interconnection, and the present invention is not limited to any particular mechanization for interconnecting the probes and console. Console 16 is typically implemented by a conventional personal computer; however, other forms of computers, such as a Sun, Hewleft Packard, or IBM Unix workstation, may also be equipped and utilized as the console in substantially the same manner described below. Specifically, console 16 may be implemented by an IBM-compatible personal computer preferably equipped with a mouse, monitor, keyboard and base. The base commonly contains the processors, memory and communications resources, such as internal/external modems or other communications cards for the console. The console includes software for analyzing the data collected by the corresponding probe or probes and displaying the information to an operator, for example, in a manner similar to that described in U.S. patent application Ser. No. 08/746,416, the disclosure of which is incorporated herein by reference in its entirety.
Further, console 16 can utilize short term and long term databases to maintain data for extended periods of time. The databases may be implemented by any conventional or commercially available database. Console 16 may operate as a stand-alone console coupled to the probes, or in a client/server configuration wherein a server (i.e., a computer system as described above preferably utilizing a windows NT environment) performs substantial interactions with the probe and conveys probe information to its clients (i.e., computer systems as described above preferably utilizing a Windows 95 NT or Unix environment). Users may also communicate with the probe directly for the data collection requests and for providing configuration parameters to the probe.
While console 16 shown in FIG. 1 as a single console, it will be understood that the present invention is not limited to a single-console embodiment, and more than one console can be used to process and display monitoring information. For example, two or more consoles at different sites or locations can be used to receive data from one or more probes. In such a plural-console arrangement, the consoles would preferably be interconnected (e.g., via the network or by other communication lines) so that each console can display or record all of the monitoring information of relevance to a related site or set of sites. Accordingly, the term "console," as used herein, can be a single unit or plural, interconnected units located at one or more sites or locations, wherein the unit(s) process, display, record and/or archive system performance data in any format, such as those disclosed in U.S. patent application Ser. No. 08/746,4160.
Console 16 receives the delta PDU counts from both probes A and B, thereby allowing console 16 to compute the two-way data delivery ratio for the VC connecting sites A and B. Specifically, DDR is calculated from the aggregate of the counts from both probes by:
Again, note that calculation of DDR by this method works even if inter-probe message PDUs are lost, because the inter-probe messages carry raw counts, and the delta PDU counts are computed by the probes themselves. The round trip delay (RTD) and DDR performance metrics can then be manipulated, recorded and displayed by the console 16 in any suitable manner to depict the SLA performance at particular times or over a period of time.
Note that, for a single transmission circuit, both of the probes at the respective ends of that circuit will generally provide the RTD measurement (i.e., two identical measurements). This redundancy allows the console to receive a valid RTD even when one of the two probes fails to receive the necessary timestamp information to calculate the RTD in a measurement cycle (e.g., when the final message PDU is lost).
By way of non-limiting example, the SLA data can be presented in single tabular report as represented in the example below. Numerous other reports can be prepared from the accumulated data.
Where the service offering describes several classes of service, each having their own service level, reports can be by service class. This can be implemented by creating each service class as a separate network. For example, all of the circuits with the SNA class of service can be configured to be in the "SNA" network. Preferably, the database stores a service class parameter for each circuit. This can be used to create reports by service class.
Optionally, the user can configure a single, weekly maintenance window of time that is automatically excluded from the SLA measurements. The window is specified by the day of the week, its start time and its duration. To automatically exclude periods of unavailability due to maintenance activity that occurs outside of a regularly scheduled maintenance window, custom integration to a trouble-ticketing system is required. Alternatively, a special type of data editor can be used to manually control exclusion of measurements during periods of unavailability due to maintenance.
While the above-described messaging sequence and collected measurement data support both round-trip delay (RTD) and data delivery ratio (DDR) network performance metrics, it will be understood that these metrics are separable and need not be determined in conjunction with each other, i.e., RTD can be determined without determining DDR, and DDR can be determined without determining RTD. Moreover, while equations (1) through (6) set forth specific expressions for determining RTD and DDR, the present invention is not limited to these specific expressions, and the timestamp and PDU count measurements can be used to calculate different or modified measures of network delay and percentages of successful (or unsuccessful) data transmission.
In accordance with another aspect of the present invention, the inter-probe message structure used to transmit inter-probe messages M.sub.1 through M.sub.4 is designed to allow each inter-probe message to be sent within a single PDU through the switching network, irrespective of the data transmission protocol(s) employed in the network, including ATM, Frame Relay, etc. This feature of the present invention advantageously permits the messaging scheme to be used with any conventional packet switching network, including interworked switching networks. This feature is achieved by minimizing the number of bytes required to send a single inter-probe message, so that the message can be encapsulated within the smallest PDUs used in the network, e.g., a single ATM cell. FIG. 4 illustrates the data payload of an inter-probe message according to the exemplary embodiment of the present invention. As seen in FIG. 4, the inter-probe message payload to be encapsulated within a PDU includes 20 bytes.
A one-byte "Version #" field identifies the product version number to which the message structure corresponds (e.g., a "2" indicates that the message structure corresponds to the second release of the product).
A one-byte "Message Type" field identifies the current state of the overall SLA message exchange, i.e., it identifies whether the message is M.sub.1, M.sub.2, M.sub.3 or M.sub.4. A value of 1 indicates that the message is the "initial request" message M.sub.1 ; a value of 2 indicates that the message is the "reply.sub.B " message M.sub.2 ; a value of 3 indicates that the message is the "reply.sub.A " message M.sub.3 ; and a value of 4 indicates that the message is the "final reply" message M.sub.4. Even if the probes on both ends of the VC begin an SLA message exchange with the same sequence number, the "Message Type" field is used to resolve the potential ambiguity.
A two-byte "Sequence #" field identifies the SLA message exchange of which the message is a part (somewhat analogous to the index "i" described above to denote the measurement cycle).
A four-byte "SlaUpTime" field is a count of seconds since the probe booted, SLA was enabled, or this particular circuit became available. The SlaUpTime of the reporting probe is included in the inter-probe message so that reboots, SLA enabling, and circuit reconfigurations can be detected and accounted for in the probes' calculations of delta PDU counts.
A four-byte "Last RX timestamp" field is the timestamp (in milliseconds and 1024ths of a millisecond) of when the probe received its last SLA message PDU. Specifically, for the reply.sub.B message M.sub.2, this field contains the timestamp indicating the value of time T.sub.2, and for the reply.sub.A message M.sub.3, this field contains the timestamp indicating the value of time T.sub.4. For messages M.sub.1 and M.sub.4, this field contains no valid data.
A four-byte "Last TX timestamp" field is the timestamp (in milliseconds and 1024ths of a millisecond) of when the probe offered its last SLA message PDU to the network. Specifically, for the reply.sub.A message M.sub.3, this field contains the timestamp indicating the value of time T.sub.1, and for final reply message M.sub.4, this field contains the timestamp indicating the value of time T.sub.3. For messages M.sub.1 and M.sub.2, this field contains no valid data.
Both timestamp fields consist of 22 bits of integer milliseconds (upper 22 bits) and 10 bits of 1024ths of a millisecond (lower 10 bits). This allows for about 69 minutes between timestamps before they wrap. Internal timestamp values are in milliseconds, but the fractional part allows for the potential of higher precision timestamps, and allows the serialization delay and internal delay factors to be combined with less accumulated round off error. The timestamps may be relative to the probe SlaUpTime.
A four-byte "non-burst offered PDUs" field is a free-running (raw) count of the number of PDUs offered to the network on the transmission circuit, recorded at the time the sending probe sent its first message in the message sequence. Specifically, for the reply.sub.A message M.sub.3, this field contains the raw count, since counter initialization/rollover, of the PDUs offered by probe A up to time T.sub.1 (the time at which probe A transmitted message M.sub.1). For the final reply message M.sub.4, this field contains the raw count, since counter initialization/rollover, of the PDUs offered by probe B up to time T.sub.3 (the time at which probe B transmitted message M.sub.2). For messages M.sub.1 and M.sub.2 in the message sequence, this field contains no valid data. The SlaUpTime is used to account for counter rollover and probe reboots.
32-bit unsigned counters can be used by probes A and B to store the PDU counts in order to ensure infrequent counter rollovers with these counts. Assuming a worst case of 128 octets frames at full DS3 (45 MBps), the 32-bit PDU counter will roll over only once every 27 hours. 45 MBps/(128*8)=44,000 fps. 2.sup.32 /44,000=97,600 secs=27 hours. If the rollover rate becomes unmanageable for higher transmission rates, this counter can be extended to 64 bits in a straight-forward manner.
The above-described message data payload is encapsulated in a PDU for transmission over the switching network. For probes operating under the DS3 ATM protocol, the size and format of the SLA message PDU is fixed. For probes operating under the Frame Relay protocol, the SLA message PDU size can be user selectable.
Specifically, for a 53 byte ATM message PDU, the PDU contains the following: a 5 byte ATM header; 8 bytes of encapsulation; 20 bytes of message payload; 12 bytes pad; and an 8 byte ATM ML5 trailer. For a Frame Relay message PDU, the PDU contains the following: 2 bytes of Frame Relay header; 8 bytes of encapsulation; 20 bytes of payload; 0 to 224 bytes of pad (to make total frame size 32 to 256 bytes); and 2 bytes of Frame Relay FCS.
As will be understood from the foregoing, a complete SLA message exchange involves 4 messages being sent: 1) probe A sends an "initial request" message M.sub.1 to probe B; 2) probe B replies with a "reply.sub.B " message M.sub.2 ; probe A replies with a "reply.sub.A " message M.sub.3 ; and probe B replies with a "final reply" message M.sub.4. The valid fields in the four messages are summarized in Table 2 (also see FIG. 2).
If at any point in an SLA message exchange, an expected message doesn't arrive within a predetermined period of time, e.g., 10 seconds, the entire SLA conversation is terminated. On the initiating side, the conversation is reinitiated. This initial request is sent three times before giving up. The sequence number for each subsequent initial request is one greater than the previous initial request sequence number.
The inter-probe messages must be uniquely identifiable from the user traffic so that they can be transparently inserted into and extracted from any legal Frame Relay, ATM or other protocol data stream.
A probe operating in the Frame Relay protocol can use RFC 1490 and SNAP ethertype encapsulation with a proprietary (e.g., Visual Networks, Inc.) SLA ethertype value for the inter-probe messages. This encapsulation method is used for SLA messages regardless of the user's network encapsulation configuration.
An exemplary eight byte RFC 1490 encapsulation header for Frame Relay (all values in Hex) is shown in Table 3.
The "Ethertype" field in the last two bytes of the SNAP header identifies a proprietary protocol type of the packet. If the user data is also encapsulated with RFC 1490, there is no problem in differentiating the user data from the probe messages. If the user data is encapsulated with Cisco/EtherType, the probe messages contain an invalid EtherType (03 00); thus, differentiation is easily achieved. Even if some other (proprietary) user data encapsulation is in use, it is highly unlikely that user data would be erroneously identified as an SLA message, because all 8 encapsulation bytes would be required to match. As a safeguard, however, a local configuration setting allows the user to disable SLA messaging.
A probe operating in the ATM protocol can use a similar technique based on RFC 1483 encapsulation. The RFC1483 header is followed by SNAP-Ethertype and a proprietary SLA Ethertype value. The resulting eight byte RFC 1483 encapsulation header for ATM (all values in Hex) is shown in Table 4.
While the above, messaging sequence protocol and message structure are particularly advantageous because of the ability to encapsulate each message within an single PDU of any conventional data transmission protocol, it will be understood that the present invention can be carried out with other message protocols and message structures. By way of example, in accordance with another embodiment of the present invention, the aforementioned time stamps and PDU counts are transmitted using an alternative messaging scheme which is particularly suited to the Frame Relay protocol.
Specifically, probe A send a first message M.sub.1 containing a timestamp of the transmit time, T.sub.1, and the count of the number of PDUs offered (OFF.sub.A (i)) by site A (bound for site B) to the network up to time T.sub.1. Since the time T.sub.1 is not precisely known until message M.sub.1 is actually transmitted, the transmitted values of T.sub.1 and OFF.sub.A (i) are estimated.
When the message M.sub.1 is received by probe B on the other end of the circuit, probe B records a timestamp of the receive time, T.sub.2, of the message M.sub.1. Probe B then sends a reply message M.sub.2 back to probeA at time T.sub.3. Message M2 includes timestamps indicating the values of times T.sub.2 and T.sub.3 as well as the count of number of PDUs offered (OFF.sub.B (i)) by site B (bound for site A) to the network up to time T.sub.3. Once again, time T.sub.3 and PDU count OFF.sub.B (i) must be estimated, since they are not known precisely until the actual transmit time T.sub.3 of the message M.sub.2. Alternatively, timestamps T.sub.1 and T.sub.3 must be taken as close as possible to the actual time the messages are transmitted so that time spent waiting for idle time (a break in the user traffic) is not included in the measurement. Upon receiving the reply message M.sub.2 at time T.sub.4, probe A can compute the round-trip delay in accordance with equation (1). This final result is then sent back to probe B in one final message M.sub.3, so that probe B can record the result as well (probe B is not sent a timestamp for time T.sub.4 and cannot, therefore, calculate RTD).
The above-described inter-probe messages conform to message format shown in Table 5.
The fields T.sub.1, T.sub.2 and T.sub.3 contain the timestamps for times T.sub.1, T.sub.2 and T.sub.3, respectively, and the RTD field contains the round trip delay time. The Measurement Type field indicates whether the message is an M.sub.1, M.sub.2 or M.sub.3 type message. The remaining fields contain information similar to the information contained in the corresponding fields described above in conjunction with FIG. 4.
As will be understood from above, a complete message exchange sequence includes three messages: 1) probe A sends an "initial request" message to probe B; 2) probe B replies with a "reply" message; and 3) probe A replies with a "result reply" message. The valid fields in the four messages are summarized in Table 6.
To accommodate service level agreements that specify delay in terms of frame sizes other than 200 bytes, the inter-probe message size can be user programmable. Having described preferred embodiments of a new and improved method and apparatus for performing service level analysis of communications network performance metrics, it is believed that other modifications, variations and changes will be suggested to those skilled in the art in view of the teachings set forth herein. It is therefore to be understood that all such variations, modifications and changes are believed to fall within the scope of the present invention as defined by the appended claims.