CN1228568A - Computer network service device - Google Patents
Computer network service device Download PDFInfo
- Publication number
- CN1228568A CN1228568A CN 99101647 CN99101647A CN1228568A CN 1228568 A CN1228568 A CN 1228568A CN 99101647 CN99101647 CN 99101647 CN 99101647 A CN99101647 A CN 99101647A CN 1228568 A CN1228568 A CN 1228568A
- Authority
- CN
- China
- Prior art keywords
- network
- machine
- file
- server
- node
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Landscapes
- Computer And Data Communications (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The present invention relates to a computer network service device. It is characterized by that several computers are connected together by means of network in the server machine box so as to make its whole body possess the characteristics of large storage capacity, high transmission speed, strong fault tolerant ability and expansible scale, and provide multimedia service for users in the presence of necessary hardware and software method.
Description
The present invention is a kind of Computer Systems Organization that multimedia service is provided for the network user.
As everyone knows, multimedia server needs the very big storage capacity and the great network bandwidth.Particularly when Video service was provided, as after compressing in the MPEG1 mode, its bandwidth was more than 1.4Mbps to video flowing, be converted to storage space, 1 hour is 1.4 * 3600/8=630M byte, is 90 minutes by a film average length, will need the space of 630 * 1.5=945M byte; As compressing in the MPEG2 mode, its bandwidth is more than the 4.2Mbps, and one hour storage space is the 1890M byte, and the storage space of a film is the 2835M byte.If server will provide a few tuber of stemona films for hundreds of users, the storage space of the network bandwidth more than the 1Gbps and hundreds of GB and even number TB will be needed.Owing to do not see the dedicated multimedia server on the market so far, according to the product of at present external producer, never outer to separating of this problem is three kinds of schemes: high performance small-size computer or even mainframe computer are joined disk array and are formed multimedia server.The typical case of this scheme representative is an IBM (IBM) based on the multimedia server (50 following users) of RS/6000 minicomputer with based on the multimedia server of ES/9000 large scale computer.Second kind of scheme is to adopt massively parallel computer to add the multimedia database server that large-scale set of databases becomes.Typical case's representative of this scheme is the multimedia server that U.S. Oracle company makes up based on U.S. n-Cube company massively parallel computer, this server has the thousands of video flowings of the concurrent processing of thousands of processors, its I/O passage is up to 1024, and the disk of support more than 6000 can walk abreast.The third scheme is to add disk array with an above minicomputer or even PC to form (Cluster) system of trooping, and making up multimedia server by special software, the typical case of this scheme is represented as the Tiger software of company of U.S. Hewlett-Packard (HP) MC/Service Guard software and Microsoft (Microsoft) company.Why adopting such scheme abroad, is because multimedia server must satisfy following requirement:
1, the disk storage space of hundreds of GB and even thousands of GB must be provided.At present the single disk that can realize of prior art adopts being to the maximum about 60GB of scsi interface, adopt ide interface about 10GB, realize aforementioned memory requirement, must adopt a plurality of disks.And on ide interface, can only connect 4 disks at most, and scsi interface connects seven at most, and the Ultra2 scsi interface can be received 15 at most.Simultaneously, but the quantity that connects disk also is subjected to the restriction of interface transmission speed and host process ability except that the restriction that is subjected to interface to connection device quantity, can not take over many disks.Here it is, and all multimedia servers all adopt the reason of scsi interface and disk array.
2, the network bandwidth of hundreds of and even thousands of Mbps must be provided.Present single network interface card maximum can provide the bandwidth of 100Mbps, and meet the demands must a plurality of network interface cards, and this has just required a plurality of slots to use for network interface card, generally has only mainframe just to have enough slots to satisfy the demand.Priorly provide so that the big network bandwidth needs main frame that enough processing poweies and input/output capabilities are arranged.Therefore, external such scheme is not to adopt mainframe or parallel processor, and the mode that adopts many calculation mechanism to become to troop exactly solves this problem.
Because disk array, parallel processing computer or multimachine cluster system are all very expensive, cause the multimedia server cost high, be unfavorable for market acceptance.
The objective of the invention is to design a kind of cost extremely low utilize the computer networking technology structure, adopt present mass-produced components and parts to construct that a kind of memory capacity is big, transmission speed is high, have certain fault-tolerant ability and can be with application need flexible microgrid computer media server, compare under the situation of equal capacity and transmission speed with the like product of present employing prior art manufacturing, cost is less than being 1/5 of external like product.
Technical scheme of the present invention is: be provided with the computing machine more than two in machine box for server, this computing machine constitutes an integral body by network and the network equipment, switch, display and keyboard and mouse each other and carries out the work as multimedia server.This computing machine includes one or more central processing unit (CPU), above internal memory, mainboard and Peripheral Interface onboard, the external unit of 16K: as disk, video card, sound card, network interface card, modulator-demodular unit, operating system; Network packet in the server contains network connection and the network equipment such as network interface card, network connection device (HUB), the network switch (SWITCH), router (ROUTER), asynchronous transfer mode ATM.What computing machine comprised in cabinet both can be respectively independently as above-mentioned key element, also can be onboard integrated, and this computing machine both can be an independent completion, also can be by the shared display of each computer motherboard, keyboard and mouse.
Further set forth content of the present invention below in conjunction with drawings and Examples:
Fig. 1 is the embodiment of the invention one structural representation;
Fig. 2 is the empty frame line of a Fig. 1 inner structure synoptic diagram;
Fig. 3 is the embodiment of the invention two structural representations;
Fig. 4 is the empty frame line of a Fig. 3 inner structure synoptic diagram.
Shown in Figure 2 as Fig. 1, the outer computing machine of frame of broken lines is client computer (9), i.e. multimedia server service object.For constituting the network of multimedia server, network comprises at least two computing machines (1) and the one or more network switching equipment, for the user provides multimedia service in the frame of broken lines.In the ordinary course of things, every computing machine has two network interface cards, and one is connected with the network switching equipment (10), as the external interface of multimedia server.Another piece constitutes the multimedia server internal network as server inner exchanging data usefulness by this part network interface card and the network switching equipment (11), under the control of software, makes many computing machines (1) collaborative work as same integral body in the server.
The diagram frame of broken lines partly is for the explanation network structure, is actually the cabinet of this part network will being packed into, and the big arrow that accompanying drawing 1 usefulness is pointed to multimedia server from frame of broken lines has been illustrated this point.Only need in force computer motherboard (2), external equipment such as disc driver, CD drive etc. and power supply (3), the network equipment (4) are waited the cabinet of packing into and customizing, by mechanical switch or electronic switch, make the shared display of polylith mainboard (5), a keyboard (6) and mouse (7), pack into software and network parameter is set according to operating system of being packed into and procotol, the special software that operation will be described below promptly constitutes a multimedia server.Because multiple motherboard (2), external equipment (12), the network equipment (4), internetwork connection mode and different operating system, procotols can be arranged, they can have multiple combination again each other, equipment in the cabinet (8) such as mainboard (2), disc driver, power supply (3) etc. both can horizontal positioned, also can vertically place, certainly place about all right, display (5), keyboard (6) and mouse (7) can be shared, can be not shared yet, all this kind will both be enumerated one by one and can not also there is no need.Therefore, the diagram server is not just schematically represented the situation of practical application.Hardware technology of the present invention is characterised in that: comprised the computing machine (1) more than two or two in a cabinet, said here computing machine (1) comprises following key element: 1 or 1 above central processing unit (CPU); The internal storage that the 16K byte is above comprises ROM (read-only memory) (ROM) and read-write memory (RAM); Mainboard and Peripheral Interface onboard; External unit such as disk, display card, sound card, modulator-demodular unit etc.; Operating system etc.These computing machines are connected with each other by network, and said here network comprises following key element: network connection comprises cable such as twisted-pair feeder, concentric cable, polycore cable etc., optical cable (containing optical fiber), printed circuit board wiring etc.; The network equipment comprises network interface card, hub (HUB), the network switch (SWITCH), router (ROUTER), asynchronous transmission mode exchange (ATM) etc., and these equipment can be independently, also can be onboard integrated; Procotol is as TCP/IP, IPX, FDDI etc.Under the control of software and coordinating, shared resource, collaborative work, externally provide service.
A complete computer system is except having the hardware as subject, software as soul also must be arranged, because multimedia server of the present invention (hereinafter to be referred as server) hardware is not limited under the specific operating system and procotol, so we provide the first method that realizes server software with natural language rather than program language below:
Resource situation according to every computing machine (hereinafter to be referred as the node machine) in the server is set up system's table of available resources, and the resource table project comprises: this machine maximum resource, this machine buffer resource, the maximum available resources of this machine, this machine current available resource.Wherein, the maximum available resources of this machine=this machine maximum resource-this machine buffer resource.The system resource table is distributed on each node machine, by one on the node machine independently monitoring resource and maintenance process maintenance.The resource operating position of this this machine of monitoring the process, be used or because failure and other reasons causes some resource can not use and take that resource is released or fault is excluded the back resource can use again the time as resource, revise resource table node machine current available resource project, to reflect the actual operating position of resource at any time.
Be logical address of server-assignment, mapping is set up in the network address of selected node machine logical address therewith, and promptly all send to this node machine address to the request of server access.Demarcating this node machine is the initial point machine, i.e. setting of other node machine and coordination will be undertaken by this machine.The selection principle of initial point machine is: if node machine performance, configuration are different, select the node machine that performance is the highest, configuration is the most complete, otherwise optional node machine.
Be configured for other node machine by the initial point machine, comprise duplicating file being set,, the priority level of node machine etc. is set initial point machine address notification node machine to the node machine.Priority represents that with positive integer digital 0 priority is the highest, and numeral is big more, and priority is low more.The priority numeral of initial point machine is 0, and the priority of other node machine determines that according to performance and configuration performance and configuration are high more, and priority is high more.The initial point machine finishes in system configuration, restart successfully after, set up the server system table of available resources, project comprises: the node machine network address, node machine maximum resource, node machine buffer resource, the maximum available resources of node machine, node machine current available resource.This table obtains to the inquiry of node machine by the initial point machine except that the project of the node machine network address.Initial point machine and come regularly to send query requests to the monitoring resource and the maintenance process of each node machine according to the value of polling interval parameter setting, the monitoring resource of node machine and maintenance process are beamed back the initial point machine with the system resource table of this machine after receiving query requests, by initial point machine update service device system table of available resources in view of the above.
By the initial point machine is the server input data file, as computer program, movie and video programs, e-book etc. all can be by Computer Storage and processing, and the information of configuration file form.These files will be stored on the disk of each node machine, by the File Directory Table unified management.By the control input process, on every node machine, deposit quantity file about equally.It deposits process numeral according to priority, deposits from low to high.I.e. supposition has a n platform node machine, is that No. 0 node machine begins to deposit first file from the priority numeral earlier, deposits second file at No. 1 node machine then, ..., be to deposit n+1 file at the n machine at last, and n+2 file is stored in the node machine No. 0, so circulation is gone down.In depositing process, the monitoring resource process of node machine is monitoring the operating position of disk resource all the time, and when disk uses when arriving maximum available resources, this node machine will not allowed and be continued to deposit file, the initial point machine turns to next node machine to deposit, until till all the node machines are all deposited completely.For guaranteeing task scheduling, every node machine all must keep certain disk size as buffer resource, as the minimum twice of depositing the file of size maximum in the file for system of the disk size of buffer resource.In the data input process, set up File Directory Table.File Directory Table comprises: title, classification, character, original medium numbering, author, relevant information and file paths such as publisher.Under the prerequisite that guarantees title and these two key projects of file path, other project can increase and decrease.File path comprises the node machine address at file place.After File Directory Table was set up, the initial point machine was distributed to it on other node machine.File Directory Table safeguarded by the initial point machine, and increase, deletion or move must be undertaken and reflected in File Directory Table by the initial point machine.File Directory Table is if any change, and the initial point machine will refresh the File Directory Table of other node machine, to guarantee consistance.File Directory Table all leaves under the same path on all node machines, and promptly except node machine address, pathname is the same.
Server address on the client computer all is set to the server logical address, when client computer when server sends the request that requires service, be actually by map addresses and receive this request to the initial point machine of server logic address, after the initial point machine is analyzed client requests, determine the file path of client requests and determine the node machine network address at file place in view of the above according to File Directory Table, obtain its resource situation according to the node machine network address again.The situation that depends on this node machine current available resource, the initial point machine is taked different actions: if this node machine current available resource is more than or equal to satisfying the client requests resource requirement, the initial point machine is issued client computer with this node machine network address, and the notice client computer connects with this address and this node machine and obtains service.If this node machine current available resource is less than satisfying the client requests resource requirement, the initial point machine is searched the node machine network address of current available resource maximum in the server system table of available resources, and relatively whether the current available resource of this node machine satisfies the client requests resource requirement, if this node machine current available resource is more than or equal to satisfying the client requests resource requirement, the initial point prow notifies this node machine by the file of internal network from the node machine download client machine request of file place earlier, then client computer is issued in this node machine network address, the notice client computer connects with this address and this node machine and obtains service, for stream file as this class of MPEG, the node machine needn't pending file all transmits and just begins to provide service after finishing, can receive file on one side, Yi Bian provide service so that in time satisfy client's needs.After file all copied to this node machine, the initial point machine increased new record to reflect this variation at file in the record table, and when the client requests identical file is arranged again, the initial point machine will have more choice when searching available resources.If this node machine current available resource is less than satisfying the client requests resource requirement, then come regularly querying server system table of available resources by the client computer wait of initial point machine notice and according to the value that the polling interval parameter is provided with, if the current available resource that certain node machine is arranged behind certain hour is more than or equal to satisfying the client requests resource requirement, have the file that client computer is asked clearly as this node machine, the initial point machine is issued client computer with this node machine network address, and the notice client computer connects with this address and this node machine and obtains service.Otherwise the initial point prow notifies this node machine by the file of internal network from the node machine download client machine request of file place earlier, then client computer is issued in this node machine network address, and the notice client computer connects with this address and this node machine and obtains service.After service finishes, client interrupt with recover again being connected of node machine and being connected of initial point machine.Meanwhile, node machine monitoring resource and maintenance process discharge occupied resource, revise system's table of available resources and it is issued the initial point machine, and the initial point machine is refreshing service device system table of available resources in view of the above.For the disk buffering resource, at first to guarantee after finishing service, must to have enough spaces can storage system in the file of size maximum, also to preserve the high file of some access frequencys as far as possible.Therefore, to the file in the disk buffering resource, node machine monitoring resource and maintenance process will be according to being determined flexible strategy the access times/unit interval of file, access times/the unit interval of file is many more, flexible strategy are big more, according to the definite file that will delete of the principle of little flexible strategy and first in first out, promptly at first delete the file of flexible strategy minimum then, if flexible strategy equate then delete the longest file of resting period in the disk buffering resource.
When system configuration, initial point machine program has been installed on each node machine, is the operation of the program start on the node machine of initial point machine but have only to demarcate.Initial point machine program is except finishing above-mentioned work, also regularly send query requests at interval according to polling interval parameter official hour to the monitoring resource and the maintenance process of every node machine by internal network, after the monitoring resource of node machine and maintenance process are received the query requests of initial point machine, by internal network this machine system table of available resources is issued the initial point machine, the initial point machine upgrades server system table of available resources thereon, to avoid distributing disabled resource.If the initial point machine is not received replying of a certain node machine in maximum latency parameter official hour, under the situation that the reserve internal network is arranged, the initial point machine starts the reserve internal network to be attempted connecting with this node machine, if can receive replying of node machine this moment, and initial point machine and node machine will carry out that inside is connected and note down network error information on journal file and wrong output device with backup network.If do not have backup network or backup network is arranged but still can't connect by backup network, the initial point machine just thinks that this node machine can not work, is designated as this node machine unavailable in File Directory Table and server system table of available resources and record node machine failure error message on journal file and wrong output device.If the monitoring resource of node machine and maintenance process are not received the query requests of initial point machine in maximum latency parameter official hour, under the situation that the reserve internal network is arranged, the node machine starts the reserve internal network to be attempted connecting with the initial point machine, if can receive replying of initial point machine this moment, and initial point machine and node machine will carry out that inside is connected and note down network error information on journal file and wrong output device with backup network.If do not have backup network or backup network is arranged but still can't connect by backup network, the node machine is attempted connecting with other node machine, if can not connect, the node machine is Minute book machine network error information on journal file and wrong output device.If can connect and other node machine can also be received the query requests of initial point machine, the node machine is connected initial point machine network error information at journal file with Minute book machine on the wrong output device.If can connect the query requests that other node machine and other node electromechanics can not be received the initial point machine, node machine medium priority numeral node machine minimum, that priority level is the highest will start the initial point machine program that has installed but do not started as yet before this, the network address of this machine is mapped to the server logic address, demarcates that this machine is carried out the work for the initial point machine and record initial point machine failure error message on journal file and wrong output device.In this way, when some part breaks down, the server available resources are reduced, whole server is quit work.After breaking down, because the node machine only by network interconnection, can be fixed a breakdown to the part that goes wrong under the situation that does not stop server work, this part can add work again after fault was got rid of under the non-stop-machine situation of server.
A representative instance of this server implementation method.
Be that a family has the deluxe hotel in 500 guest rooms to make up a guest room network.This network also will provide the service of movie and video programs program request to the guest except sending and receiving e-mail being arranged, carrying out the function of word processing.Requirement can provide the VCD about three tubers of stemona movie program, satisfies the needs of house rate whole guests program request simultaneously under 100% situation.Realize this requirement, server must have the above disk size of 300GB, the transmission speed that 800Mbps is above.
11 mainboards that use Intel 440BX chipset are installed in machine box for server, and central processing unit is the Pentium II 450MHz CPU of Intel Company.Save as 128MB in the every mainboard, join an AGP display card, the network interface card of 3 100Mbps.Wherein 10 mainboards are respectively joined the hard disk that 4 10GB are with the 256KB high-speed caches, adopt Ultra DMA/66EIDE interface, another piece mainboard is joined two 10GB band 256KB high-speed caches, is adopted the hard disk of Ultra DMA/66EIDE interface, and order is joined a DVD CD drive, the magnetic tape station of a 40GB.Adopt two fast ethernet switch that 24 100Mbps mouths are arranged as the network switching equipment, wherein one with every mainboard on a network interface card be connected by super category 5 twisted pair, sharedly remove 11 mouths, also have 13 confessions to be connected usefulness with the guest room net, the effect of this switch is shown in the network switching equipment 1 in the accompanying drawing 1.Another switch is connected with other two network interface cards on the every mainboard, and as internal network and backup network usefulness, the effect of this switch is shown in the network switching equipment 2 in the accompanying drawing 1.Server uses a display, keyboard and mouse, is connected with display card, keyboard interface, the mouse interface of every mainboard by divider.Every mainboard is powered with one 300 watts power supply.According to above-mentioned configuration, server is become by 11 joint mechanisms, the hard-disk capacity of 480GB is arranged, the internal memory of 1.25GB.
Every node machine is installed (SuSE) Linux OS, and operating system and application software etc. comprises that the exchange area takies the 500M hard drive space altogether.Keep 9GB as the disk buffering resource, also have the 30GB space to can be used for the store data file.Procotol adopts TCP/IP.To there be that node machine of CD drive and magnetic tape station to demarcate for the initial point machine, according to other node machine of setting noted earlier.By CD drive and the magnetic tape station program of packing into, set up File Directory Table.The client software of packing on client computer, system can move.
This station server has the above free space of 300GB, the inbound bandwidth of 100Mbps, the departures bandwidth of 1100Mbps.When internal network switch and network interface card all are made as full duplex mode, the inner exchanging bandwidth of 200Mbps can be provided, as backup network is also come into operation, can provide the inner exchanging bandwidth of 400Mbps.
As whole users during, will need the departures bandwidth of 500 * 1.4Mbps=700Mbps all at the movies on demand program.Each user's services request can not surpass 100Kbps, will need 500 * 100Kbps=50Mbps.The network bandwidth that server provides can be satisfied the demand fully.
Except the phase-split network bandwidth, also must analyze the input/output capabilities of disk.The extreme case that produces disk input/output capabilities bottleneck occurs in works as whole users all at same movie program of program request, and the program request time has nothing in common with each other but concentrates in the time span of this program, simultaneously, and when the file of this program does not copy to other main frame as yet.In this case, by aforementioned task scheduling mode, load at first concentrates on the node machine at this program file place, and the 100M network interface card of this node machine can be supported 100/1.4=71 user theoretically.The hard disk maximum transfer speed of system configuration is 66MB/s, continue transmission speed at 20MB/s, and 100Mbps/8=12.5MB, that is to say, satisfying under the network transfer speeds that this machine supports, hard disk also has the ability of 7.5MB/s, can with the speed of 1MB/s by internal network with file copy to other seven node machines, every node machine can be supported 71 users again again, can support 8 * 71=568 user altogether.In fact, just can meet the demands as long as xcopy is not less than 1.4Mbps to the speed of other node machine, by 7 * 1.5/8=1.3125MB, hard disk only need can meet the demands less than the lasting transmission speed of 14MB/s.Even if every node machine can not be supported 71 users, this will make hard disk can distribute the more bandwidth resource to be used for xcopy to more node machine, thereby have more node machine that service is provided, the input/output capabilities of this proof disk can be satisfied the demand.
The multimedia server of external import is under the situation of identical configuration and performance index, and its price can not be lower than 500,000 dollars, and multimedia server price of the present invention can not surpass 500,000 Renminbi.Owing to adopt mass-produced tested parts, add fault-tolerant design simultaneously, its quality can be than imported product difference yet.
Second kind of hardware implementation method of the present invention can be illustrated with Fig. 3 and Fig. 4:
The outer equipment of frame of broken lines is general television set and stereo set, be connected with multimedia server by video signal cable, audio signal wire and control line, wherein the contract TV output port of card of video signal cable and display card or mpeg decompression is connected, the contract audio output port of card of audio signal wire and sound card or mpeg decompression is connected, control line connects an infrared ray or radio system at an end of televisor, the other end is connected with the RS232 serial port, and the user can operate with telepilot or wireless mouse equal controller.For constituting the network of multimedia server, network comprises at least two computing machines and a network switching equipment in the frame of broken lines.In the ordinary course of things, every computing machine has a display card, a sound card, RS232 serial port and the mpeg decompression that also can the not have card that contracts can be arranged is connected with televisor, stereo set and control device by video signal cable, audio signal wire and control line.Every computing machine also must have a network interface card, use as server inner exchanging data, constitute the multimedia server internal network by this part network interface card and the network switching equipment, under the control of software, make many computing machines collaborative work as same integral body in the server.
The diagram frame of broken lines partly is for the explanation network structure, is actually the cabinet of this part network will being packed into, and the big arrow that accompanying drawing 2 usefulness are pointed to multimedia server from frame of broken lines has been illustrated this point.Only need in force the cabinet of the customizations of packing into such as computer motherboard, external equipment such as disc driver, CD drive etc. and power supply, the network equipment, by mechanical switch or electronic switch, make the shared display of polylith mainboard, keyboard and mouse, pack into software and network parameter is set according to operating system of being packed into and procotol, the special software that operation will be described below promptly constitutes a multimedia server.Because multiple motherboard, external equipment, the network equipment, internetwork connection mode and different operating system, procotols can be arranged, they can have multiple combination again each other, equipment in the cabinet such as mainboard, disc driver, power supply etc. both can horizontal positioned, also can vertically place, certainly place about all right, display, keyboard and mouse can be shared, also can be not shared, all this kind will both be enumerated one by one and can not also there is no need.Therefore, the diagram server is not just schematically represented the situation of practical application.The hardware technology feature of this method is identical with first method, no longer repeats herein.
Provide the second method that realizes server software below:
The central idea of this method is that server also is a client computer simultaneously.Every computing machine (hereinafter to be referred as the node machine) in the server itself is a client computer, is again the part of server.
Write a shell (Shell) program on the operating system for the reality use, can call the program of appointment, simultaneously, have only, just can withdraw from this shell, with arriving operating system by behind the keyboard input password by this shell.
The hard disk of node machine is divided into three parts such as components of system as directed, buffer portion, data file portion, and wherein components of system as directed is put operating system and application program, generally speaking, is read-only and can not be shared.Buffer portion can be read and write but can not be shared.Data file portion must be shared, and can read but must just can write under the situation of authorizing.
Selected node machine is the initial point machine.The selection principle of initial point machine is: if the configuration of node machine is different, select the most full node machine of configuration, otherwise optional node machine.All node machine hard disc data file part initial point machines can see and have it to write password, therefore can be the server input data file by the initial point machine, as computer program, movie and video programs, e-book etc. all can be by the information of Computer Storage and processing and configuration file form.These files will be stored on the disk of each node machine, by the File Directory Table unified management.Different with first method, the initial point machine is the store data file not, and son is born management role, by the control input process, deposits quantity file about equally on every node machine.It is deposited process and can deposit from low to high by numbering.I.e. supposition has a n platform node machine, begins to deposit first file from the node machine that is numbered No. 1 earlier, deposits second file at No. 2 node machines then, ..., be to deposit n file at the n machine at last, and n+1 file is stored in the node machine No. 1, so circulation is gone down.In depositing process, when the data file portion of node machine hard disk is piled, this node machine will not allow and continue to deposit file, and the initial point machine turns to next node machine to deposit, until all the node machines all deposit full till.In the data input process, set up File Directory Table.File Directory Table comprises: title, classification, character, original medium numbering, author, relevant information and file paths such as publisher.Under the prerequisite that guarantees title and these two key projects of file path, other project can increase and decrease.File path comprises the node machine address at file place.After File Directory Table was set up, the initial point machine was distributed to it on other node machine.File Directory Table safeguarded by the initial point machine, and increase, deletion or move must be undertaken and reflected in File Directory Table by the initial point machine.File Directory Table is if any change, and the initial point machine will refresh the File Directory Table of other node machine, to guarantee consistance.File Directory Table all leaves under the same path on all node machines, and promptly except node machine address, pathname is the same.
When work, after the client selectes a file by shell from File Directory Table, shell is at first determined the position of this document by file path, if this document is not on this machine, then start file that a thread requires from file place node machine download client by internal network buffer portion to this machine hard disk, and file produces a new path for this reason, again the environmental variance file path is put in this path.If this document on this machine, is then directly put into the environmental variance file path with its path.Shell is then from its program that need call of type decision of this document, when calling this program with the environmental variance file path as parameter, thereby the Actual path of file is passed to program, satisfy requirement of client.For the stream file as this class of MPEG, the node machine needn't pending file just begins to provide service after all transmission is finished, on one side can receive file, provides on one side and serves so that in time satisfy client's needs.After file all copied to this node machine, node machine notice initial point machine increased new record to reflect this variation in File Directory Table, and the initial point machine refreshes the File Directory Table of all node machines more in view of the above.For the hard disk buffer portion, at first to guarantee after finishing service, must to have enough spaces can storage system in the file of size maximum, also to preserve the high file of some access frequencys as far as possible.Therefore, file to the disk buffering part, node machine shell will be according to being determined flexible strategy the access times/unit interval of file, access times/the unit interval of file is many more, flexible strategy are big more, according to the definite file that will delete of the principle of little flexible strategy and first in first out, promptly at first delete the file of flexible strategy minimum then, delete the longest file of resting period in the disk buffering part if flexible strategy equate.
Initial point machine program is except finishing above-mentioned work, also regularly send query requests at interval according to polling interval parameter official hour to the shell of every node machine by internal network, after the shell of node machine is received the query requests of initial point machine, reply the initial point machine by internal network.If the initial point machine is not received replying of a certain node machine in maximum latency parameter official hour, under the situation that the reserve internal network is arranged, the initial point machine starts the reserve internal network to be attempted connecting with this node machine, if can receive replying of node machine this moment, and initial point machine and node machine will carry out that inside is connected and note down network error information on journal file and wrong output device with backup network.If do not have backup network or backup network is arranged but still can't connect by backup network, the initial point machine just thinks that this node machine can not work, in File Directory Table this node machine is designated as unavailablely, in view of the above the File Directory Table of all node machines is refreshed and record node machine failure error message on journal file and wrong output device simultaneously.In this way, when some part breaks down, the server available resources are reduced, whole server is quit work.After breaking down, because the node machine only by network interconnection, can be fixed a breakdown to the part that goes wrong under the situation that does not stop server work, this part can add work again after fault was got rid of under the non-stop-machine situation of server.
A representative instance of second kind of implementation method of server.
Be Play System of nightclub's design that there are ten karaoke chambers in a family, require each 2,000 head of MTV program of energy while DVD playing back and VCD quality, do not change existing device and can increase expansion equipment according to the business development situation from now on.According to these requirements, realize that server is as follows:
11 mainboards that use Intel 430TX chipset are installed in machine box for server, and central processing unit is the Pentium 200MMX CPU of Intel Company.Save as 32MB in the every mainboard, join a PCI display card, the network interface card of a 10Mbps.Wherein 10 mainboards are respectively joined the hard disk that 4 10GB are with the 256KB high-speed caches, adopt Ultra DMA/33EIDE interface, a PCI sound card, one can be to the decompression card of MPEGI and the decompression of MPEG II, another piece mainboard is joined the hard disk of a 4GB ide interface, and joins a DVD CD drive.Adopt an Ethernet switch that 12 10Mbps mouths are arranged as internal network usefulness, the effect of this switch is shown in the network switching equipment in the accompanying drawing 2.Server uses a display, keyboard and mouse, is connected with display card, keyboard interface, the mouse interface of every mainboard by divider.The televisor of private room is by video signal cable and decompression card connection, and the Karaoke power amplifier is connected with sound card by audio signal wire, and the telepilot receiving trap is connected on the RS232 mouth on the corresponding mainboard by signal wire.Every mainboard is powered with one 300 watts power supply.According to above-mentioned configuration, server is become by 11 joint mechanisms, the hard-disk capacity of 404GB is arranged, the internal memory of 350MB.With the node machine of joining 4GB hard disk and DVD driver as the initial point machine.
All node machines all adopt Windows98 as operating system, and except that the initial point machine, the hard disk partition of node machine is as follows: components of system as directed 1GB, buffer portion 3GB, data file portion 36GB.The file storage of the total 360GB of whole server can satisfy each 2,000 first above requirement of MTV program of depositing DVD and VCD quality, as only depositing the MTV program of VCD quality, can deposit more than 10000 head.Every client computer all is furnished with the card of decompression simultaneously, can DVD playing back and VCD program.Substantially do not need increase equipment in the private room, and can expand at any time on request.The investment of this scheme is no more than 250,000 yuan of Renminbi.
In a word, these two kinds of methods can make up the multi-media server system at different objects, different purposes, also the two can be combined, and satisfy the needs of different situations.
The present invention realizes by the combination setting of some computers of direct employing compared with prior art Have on its function that memory capacity is big, transmission speed is high, fault-tolerant ability is strong, the telescopic advantage of scale, gram Taken people's custom and thought the computer in big capacity storage space, hardware facility institute shape that must large buffer memory The mistaken ideas that the computer service network that becomes is realized. The present invention is real with the cost that is lower than same kind of products at abroad 1/5 Showed its all service functions that should possess, made the microgrid computer media server can be with rational cost Enter field and the places such as vast community, enterprise, hotel, for society provides more colourful service.
Claims (5)
1, a kind of computer network service device, form by computer motherboard (2), hard disk drive, CD drive, power supply (3), the network equipment (4), be characterized in: the computing machine (1) more than two is arranged by network settings in book server cabinet (8), and this computing machine constitutes an integral body by network and the network equipment (4), switch, display (5) and keyboard (6) and mouse (7) each other and carries out the work as multimedia server.
2, device according to claim 1 is characterized in that computing machine (1) should comprise following key element: one or more central processing units (CPU); The internal storage that the 16K byte is above; Mainboard and Peripheral Interface onboard; External unit such as disk, display card, sound card, network interface card, modulator-demodular unit; Operating system.
3, device according to claim 1, it is characterized in that the network packet in the server contains: network connection, the network equipment (4) comprise network interface card, network integration device (HUB) network switch (SWITCH), router (ROUTER), asynchronous transmission mode exchange (ATM).
4, device according to claim 1 and 2 is characterized in that: the key element as claimed in claim 2 that computing machine (1) is comprised both can be respectively independently, also can be integrated on the mainboard (2).
5, device according to claim 1, the computing machine (1) that it is characterized in that in cabinet being connected promptly can be the computing machine of independent completion, also can be by the shared display of each computer motherboard (5), a keyboard (6) and mouse (7).
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB991016475A CN1138224C (en) | 1999-02-07 | 1999-02-07 | Computer network service device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB991016475A CN1138224C (en) | 1999-02-07 | 1999-02-07 | Computer network service device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1228568A true CN1228568A (en) | 1999-09-15 |
| CN1138224C CN1138224C (en) | 2004-02-11 |
Family
ID=5270579
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB991016475A Expired - Fee Related CN1138224C (en) | 1999-02-07 | 1999-02-07 | Computer network service device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN1138224C (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102708719A (en) * | 2012-05-21 | 2012-10-03 | 青岛港（集团）有限公司 | Simulating training equipment for ship unloader |
| CN104484014A (en) * | 2014-12-10 | 2015-04-01 | 成都思邦力克科技有限公司 | Network server |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101726639B (en) * | 2009-12-18 | 2012-06-27 | 北京世纪互联宽带数据中心有限公司 | Rack-mounted heat load and application thereof |
-
1999
- 1999-02-07 CN CNB991016475A patent/CN1138224C/en not_active Expired - Fee Related
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102708719A (en) * | 2012-05-21 | 2012-10-03 | 青岛港（集团）有限公司 | Simulating training equipment for ship unloader |
| CN104484014A (en) * | 2014-12-10 | 2015-04-01 | 成都思邦力克科技有限公司 | Network server |
Also Published As
| Publication number | Publication date |
|---|---|
| CN1138224C (en) | 2004-02-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Yu et al. | Grouped sweeping scheduling for DASD-based multimedia storage management | |
| US7010657B2 (en) | Avoiding deadlock between storage assignments by devices in a network | |
| US7216148B2 (en) | Storage system having a plurality of controllers | |
| US6298386B1 (en) | Network file server having a message collector queue for connection and connectionless oriented protocols | |
| US7287180B1 (en) | Hardware independent hierarchical cluster of heterogeneous media servers using a hierarchical command beat protocol to synchronize distributed parallel computing systems and employing a virtual dynamic network topology for distributed parallel computing system | |
| US5944789A (en) | Network file server maintaining local caches of file directory information in data mover computers | |
| US7275050B2 (en) | Storage system, a method of file data backup and method of copying of file data | |
| JP3560211B2 (en) | System and method for distributing digital data on demand | |
| US7039784B1 (en) | Video distribution system using dynamic disk load balancing with variable sub-segmenting | |
| US8762480B2 (en) | Client, brokerage server and method for providing cloud storage | |
| US9449019B2 (en) | Peer-to-peer redundant file server system and methods | |
| US5996025A (en) | Network transparent access framework for multimedia serving | |
| JP2012504262A (en) | Distributed cache placement | |
| WO2002087236A1 (en) | System and method for retrieving and storing multimedia data | |
| JP2000512416A (en) | Virtual tape data storage subsystem attached to network | |
| FI116167B (en) | Archive file server | |
| US7240143B1 (en) | Data access and address translation for retrieval of data amongst multiple interconnected access nodes | |
| CN112416892A (en) | Emergency video data cloud storage system | |
| CN107346209B (en) | Multi-disk aggregation type data storage system and implementation method and application method thereof | |
| CN1138224C (en) | Computer network service device | |
| CN1416062A (en) | Method for gathering resources of storage media on computer network | |
| Duan et al. | A high‐performance distributed file system for large‐scale concurrent HD video streams | |
| CN1902620B (en) | Virtual file system | |
| Shafer | A storage architecture for data-intensive computing | |
| Christodoulakis et al. | The KYDONIA multimedia information server |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C06 | Publication | ||
| PB01 | Publication | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| C17 | Cessation of patent right | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20040211 |