JP2007243732A - Video server system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a video server system in which a fault of a video server is accurately detected, an alarm is communicated with high responsiveness and system switching can be performed in an appropriate timing. <P>SOLUTION: If a fault occurs, a point of time at which a constant streaming function of a video server 11a, 11b is broken is predicted by a thread for monitoring the remaining capacity of a buffer at a spot such as a storage or decoder connected to a video server 11a, 11n corresponding to an input/output if watched from the server, an alarm monitoring device 16 outputs an alarm in an appropriate timing, and the video servers 11a, 11b are switched in an optimal timing. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a video server system having a function of monitoring a failure for an input / output buffer of a video server.

  In general, a video server manages its own resources such as CPU, memory, and storage, and provides a constant streaming function in response to a transmission request from a user. For a transmission request that has been accepted once, it is necessary to guarantee transmission in all aspects, and high reliability is required for its realization.

  The configuration is a combination of a host computer including a disk array, MPEG decoder, and CPU memory. In recent years, there are many combinations of general components on the market in order to build inexpensively. In such a configuration, in order to guarantee the reliability required for a video server particularly used in a broadcasting station system, a multiplexing system configuration is often adopted. To make (To Repair) as close to 0 as possible, the point is how to detect failures for each single system. In other words, it is necessary to have a mechanism that can detect the occurrence of a failure without excess or deficiency and at high speed and can take appropriate measures such as switching the system.

  FIG. 6 is a configuration diagram of a duplex video server system. The duplicated video servers 11a and 11b are composed of servers 12a and 12b, decoders 13a and 13b, encoders 14a and 14b, and disk arrays (storage) 15a and 15b, respectively, and an alarm monitoring device 16 monitors the video servers 11a and 11b. When a failure occurs in either of them, the switcher 17 switches to a healthy video server 11a, 11b, and the baseband output 18 continues streaming.

  In this case, the occurrence of a failure is detected by detecting a failure of a different method for each component such as decoders 13a and 13b, encoders 14a and 14b, and disk arrays 15a and 15b used to construct the video servers 11a and 11b. The alarm is output to the alarm monitoring device 16 via the control network 19.

Here, as a video server system, there is one in which digital video data is efficiently transmitted from a video server to a client while ensuring reliability of data transmission (see Patent Document 1).
JP 2000-358232 A

  However, an alarm provided by a component in a conventional video server system is often not defined for a system with a high real-time property such as a video server that performs a predetermined operation at a predetermined timing. It just tells you what has already happened.

  Therefore, just collecting these alarm information does not properly report the failure per single system of the video server, but it becomes a false alarm or an alarm that is too late, and can be used as it is to create a duplex system that minimizes MTTR. It is difficult to build.

  An object of the present invention is to provide a video server system having high stability as a duplex system by accurately detecting a failure of a video server, communicating an alarm with high responsiveness, and switching a system at an appropriate timing. That is.

  In the video server system according to the first aspect of the present invention, a plurality of video servers that provide a constant streaming function in response to a transmission request from a user are arranged in parallel, and in response to an abnormality notification alarm from the video server, When a failure occurs in a video server system in which the output stream is switched by an alarm monitoring device, a thread that monitors the remaining buffer capacity at the location corresponding to input / output when viewed from a server such as a storage or decoder connected to the video server It is characterized by predicting at which point the constant streaming function of the video server fails, outputting an alarm at an appropriate timing, and switching the video server at an optimal timing.

  The video server system according to a second aspect of the invention is the video server system according to the first aspect of the invention, wherein the thread that monitors the remaining amount of the buffer outputs an alarm when it takes a certain time or more to process one cycle of the thread. A watchdog thread having a function of

  A video server system according to a third aspect of the present invention is the video server system according to the first or second aspect, wherein the video server periodically transmits a heartbeat for confirming survival at a predetermined interval, When it is determined that the video server cannot be received, the target video server is determined to be in an abnormal state, and an alarm monitoring device is provided for switching to a standby video server.

  According to the present invention, in the component at the entrance / exit of the video beam system, a thread for monitoring the buffer provided for absorbing fluctuations is provided, and the remaining amount of the buffer is monitored. It can predict when the constant streaming function of the video server will break down when it occurs, and output an alarm at an appropriate timing. Therefore, the system can be switched at the optimum timing.

(First embodiment)
FIG. 1 is a configuration diagram of a video server system according to the first embodiment of the present invention. The video servers 11a and 11b that provide a constant streaming function in response to a transmission request from the user are duplicated, and the availability is improved. The video servers 11a and 11b are composed of servers 12a and 12b, decoders 13a and 13b, encoders 14a and 14b, and disk arrays (storage) 15a and 15b, respectively. The servers 12a and 12b have buffer monitoring means 20a and 20b. ing.

  The alarm monitoring device 16 detects a failure based on the information obtained from the buffer monitoring means 20a, 20b via the control network 19, and when a failure occurs in any of the alarm monitoring device 16, the switch 17 performs a sound video server 11a, Switch to 11b and continue streaming with baseband output 18. In other words, the alarm monitoring device 16 switches the output stream from the video servers 11a and 11b according to the abnormality notification alarm from the video servers 11a and 11b.

  The buffer monitoring means 20a and 20b are input side buffers (buffers 21a and 21b) corresponding to input / output when viewed from the servers 12a and 12b such as storages 15a and 15b and decoders 13a and 13b connected to the video servers 11a and 11b. The failure is monitored by a thread for monitoring the remaining amount of the output side buffers (buffers 22a and 22b).

Then, by predicting at which point the constant streaming function of the video servers 11a and 11b fails when a failure occurs, and outputting an alarm to the alarm monitoring device 16 via the control network 19 at an appropriate timing, Switch the system at the optimal timing.

  For example, monitoring of the buffers 22a and 22b (buffers on the output side) of the decoders 13a and 13b is performed as follows. FIG. 2 is an explanatory diagram for monitoring the number of empty buffers in the buffers 22a and 22b of the decoders 13a and 13b. The decoders 13a and 13b corresponding to the output units of the video servers 11a and 11b refer to the currently stored buffer amount and predict the occurrence of a failure based on the buffer amount. The number of empty buffers is monitored by monitoring the transition of the number of empty buffers during decoder playback.

  As shown in FIG. 2, a range of the number of free buffers that are regarded as abnormal is set in advance. A threshold value L0 for determining an abnormality and a maximum buffer amount L1 of the decoders 13a and 13b are set in advance, and the transition of the number of free buffers during reproduction that changes with time is measured. In FIG. 2, M1 is a buffer amount corresponding to the time required for switching when the video server 11a, 11b is switched by the switcher 17, and M2 is a buffer amount corresponding to the reproduction time with the remaining buffer capacity.

  The free buffer amount is periodically measured at time points t1, t2, t3, and t4, and it is checked whether the measured number of free buffers exceeds a set threshold value. If the threshold value is exceeded, it is determined that there is an abnormality and an alarm is output. In FIG. 2, an alarm is output at time t4 when the threshold value L0 is exceeded. With this function, it is possible to detect an abnormality in the buffers 22a and 22b of the decoders 13a and 13b before the remaining capacity of the decoders 13a and 13b is exhausted and stop the reproduction, and to switch the effective system.

  As the range of the number of free buffers to be determined as abnormal, the upper limit is the maximum number of buffers of the decoders 13a and 13b, and the difference between the upper limit and the lower limit is set to a buffer amount M1 or more that satisfies the time required until the system switching described above. Detect abnormalities as early as possible.

  Next, monitoring of the amount of increase in the free buffer of the decoders 13a and 13b will be described. FIG. 3 is an explanatory diagram for monitoring the increase in the number of empty buffers in the buffers 22a and 22b of the decoders 13a and 13b. The increase of the free buffer is monitored by monitoring the transition of the increase of the free buffer per unit time during the reproduction of the decoders 13a and 13b. For example, by setting the number of times to be a threshold for the amount of increase in the free buffer per unit time and how many times the increase exceeding that fixed amount will be abnormal, set the threshold number of times, and change the increase in the amount of free buffer being played Measure. Then, it is determined whether or not the increase amount measured at that time exceeds the threshold increase amount, and how many times the increase has occurred continuously. If the condition is satisfied, an abnormality is determined and an alarm is output.

  Now, if a certain amount of increase in the free buffer per unit time is ΔL0 and the threshold of the number of consecutive times is 3, the free buffer per unit time from time t13 to time t14 as shown in FIG. The amount of increase exceeds the fixed amount ΔL0 (first time), the increase amount of the empty buffer per unit time from the next time point t14 to the time point t15 also exceeds the fixed amount ΔL0 (second time), and the time point Since the amount of increase in the free buffer per unit time from t15 to time t16 also exceeds a certain amount ΔL0 (third time), an alarm is generated at time t16.

  The maximum value of the increase amount of the empty buffer per unit time is a value when reproduction is performed in a state where no data is supplied to the buffers 13a and 13b. For this reason, it is set to be abnormal when the empty buffer increases by an amount close to this.

  The monitoring of the buffers 21a and 21b of the reading units of the disk arrays 15a and 15b which are the input side buffers is also performed by monitoring the prediction of failure and alarming in the same manner as the buffers 22a and 22b of the decoders 13a and 13b which are the output units. Issue.

  Further, the internal buffers of the decoders 13a and 13b require an interface that can acquire the state of the decoders 13a and 13b. However, for the read portion from the disk arrays 15a and 15b, the applications in the servers 12a and 12b are required. There is an advantage that it depends on only and has versatility. In general, a large buffer is often provided in the read portion from the disk arrays 15a and 15b, and there is an advantage that advanced processing can be performed until the stream breaks down when an abnormality is detected.

  As described above, in the first embodiment, with respect to a buffer provided to absorb fluctuations in components at the entrance and exit of the video system, for example, the decoders 13a and 13b and the disk arrays 15a and 15b, The buffer monitoring means 20a, 20b performs monitoring based on the thread. In this case, the remaining capacity that is the threshold for alarm output is the capacity required to absorb the time required to transmit the alarm, the time required to switch the system, and the time including the margin. . That is, the remaining capacity serving as a threshold for outputting an alarm is determined based on the sum of these times multiplied by the bit rate of the stream.

  According to the first embodiment, the remaining amount of buffer at the location corresponding to input / output when viewed from the servers 20a and 20b such as the disk arrays 15a and 15b and the decoders 13a and 13b connected to the video servers 11a and 11b is monitored. By predicting at which point the constant streaming function of the video server fails when a failure occurs and outputting an alarm at an appropriate timing, the system can be switched at an optimal timing.

(Second Embodiment)
FIG. 4 is a block diagram of a video server system according to the second embodiment of the present invention. In the second embodiment, timer monitoring means 23a and 23b are provided in the servers 12a and 12b, compared to the first embodiment shown in FIG. The same elements as those in FIG. 1 are denoted by the same reference numerals, and redundant description is omitted.

  As shown in FIG. 4, the thread that monitors the remaining amount of the buffer is a watchdog thread that has a function of outputting an alarm when the processing of one cycle of the thread takes a certain time or more. The timer monitoring means 23a, 23b includes a watchdog thread having a function of outputting an alarm when a certain period of time is required for processing of one cycle of the thread, and the buffers 21 of the input / output units of the servers 12a, 12b, It is monitored that the operation of a thread for monitoring the remaining amount of 22 is being performed correctly.

  According to the second embodiment, since the operation of a thread or the like for monitoring the remaining capacity of the buffers 21 and 22 in the input / output unit is monitored correctly, the frequency of non-detection of abnormal states is reduced. Can do.

(Third embodiment)
FIG. 5 is a block diagram of a video server system according to the third embodiment of the present invention. In the third embodiment, heartbeat means 24a and 24b are provided in the servers 12a and 12b in the first embodiment shown in FIG. The same elements as those in FIG. 1 are denoted by the same reference numerals, and redundant description is omitted.

  Heartbeats are transmitted from the heartbeat means 24a, 24b to the alarm monitoring device 16 via the control network 19 at regular intervals at regular intervals. When the alarm monitoring device 16 determines that the heartbeat cannot be received, it determines that the target video server 11 is in an abnormal state and switches to the standby video server 11.

  According to the third embodiment, it is possible to prevent a failure from being detected due to failure to correctly receive all alarms when a failure occurs due to a failure such as an alarm communication path due to a heartbeat. The frequency of detection can be reduced.

1 is a configuration diagram of a video server system according to a first embodiment of the present invention. Explanatory drawing of the monitoring of the number of empty buffers of the buffer of the decoder in the 1st Embodiment of this invention. Explanatory drawing of monitoring the increase amount of the free buffer number of the buffer of the decoder in the 1st Embodiment of this invention. The block diagram of the video server system concerning the 2nd Embodiment of this invention. The block diagram of the video server system concerning the 3rd Embodiment of this invention. 1 is a configuration diagram of a duplex video server system. FIG.

Explanation of symbols

DESCRIPTION OF SYMBOLS 11 ... Video server, 12 ... Server, 13 ... Decoder, 14 ... Encoder, 15 ... Disk array, 16 ... Alarm monitoring device, 17 ... Switch, 18 ... Baseband output, 19 ... Control network, 20 ... Buffer monitoring means , 21 ... buffer, 22 ... buffer, 23 ... timer monitoring means, 24 ... heartbeat means

Claims (3)

  A video in which a plurality of video servers providing a constant streaming function according to a transmission request from a user are arranged in parallel, and an output stream from the video server is switched by an alarm monitoring device in response to an abnormality notification alarm from the video server In the server system, the constant streaming function of the video server breaks down when a failure occurs due to a thread that monitors the remaining buffer capacity at the location corresponding to input / output when viewed from the server such as storage and decoder connected to the video server A video server system characterized by predicting whether to perform, outputting an alarm at an appropriate timing, and switching the video server at an optimal timing.   2. The video according to claim 1, wherein the thread for monitoring the remaining amount of the buffer is a watchdog thread having a function of outputting an alarm when processing of one cycle of the thread takes a certain time or more. Server system.   The video server periodically transmits a heartbeat for survival confirmation at regular intervals. When it is determined that the heartbeat cannot be received, the target video server is determined to be in an abnormal state. The video server system according to claim 1, further comprising an alarm monitoring device that performs switching to the video server.

Images