CN101242288B

CN101242288B - A realization method for variant duplication

Info

Publication number: CN101242288B
Application number: CN200710003580A
Authority: CN
Inventors: 王军
Original assignee: WUHAN HUIGUI TECHNOLOGY Co Ltd
Current assignee: WUHAN OS-EASY CLOUD COMPUTING CO., LTD.
Priority date: 2007-02-09
Filing date: 2007-02-09
Publication date: 2010-05-19
Anticipated expiration: 2027-02-09
Also published as: CN101242288A

Abstract

The invention relates to a technology field of computer system maintenance, especially to a method for actualizing variable copy on the net, and is mainly used to install a new software in the machineroom simply and rapidly. The method is to monitor disc change caused by write operation of the disc of the computer in the machine room, then to collect the change areas of all receivers and the senders when network copying, and to send the data in the changed areas to a receiving terminal from a sending terminal. The method, compared with the conventional incremental copy method, is more rapid and convenient, and has a great advantage.

Description

A kind of implementation method of variant duplication

Technical field

The invention belongs to the computer system maintenance technical field, be specifically related to a kind of method that on network, realizes variant duplication, be mainly used in machine room, simply dispose fast new software.

Background technology

The major function of room management is divided into system protection and network to copying two parts.The anti-locking system of system protection is destroyed, and network-copy can be unified to install to all machines of whole equipment room.

● network-copy

Suppose a machine room have A1, A2,,,,,, An platform machinery requirement safeguards (we represent this n platform machine with set A), the software and hardware configuration of this n platform machine requires in full accord, the hardware configuration unanimity only needs the same machine of buying to get final product, and will be the very large thing of job amount but each machine carries out the installation of system and application software separately.For head it off, can be earlier from set A, appoint and get a machine An and install needed system and application software earlier; According to the file format of An machine institute assembling system (commonly used have FAT16/FAT32/NTFS) valid data in the disk of An machine are detected then; Next, these valid data among the machine An are sent in the set A other n-1 platform machine.So just can effectively reduce the workload of system and Software deployment in the machine room, these technology are commonly referred to as network-copy.

● the increment copy

The machine room A that supposes the front has installed the system of the n platform machine in the machine room and application software by network-copy and has configured.But having passed through a period of time has some new softwares to install again or intrasystem configuration needs to change, and such software is installed and system configuration changes and generally can not bring lot of data to increase and change.If at this time use network-copy then need to copy again valid data in the DISK to Image, yet there are most data in fact in a preceding network-copy, to finish, because the valid data in the hard disk are many, indefinite to G up to a hundred from several G, the primary network copy can need dozens of minutes to arrive the time of several hrs like this.Can be only the software that increases newly and configuration variation to be sent to all machines be an effective way in order to address this problem.But, though, be not just the same, be exactly different such as the IP address of each machine because the configuration of the n platform machine among the A is similar, this has brought the problem of some bad solutions.Below we introduce the performing step of increment copy:

1) be that transmitter is carried out network-copy to machine room A with An.Suppose that now the valid data set in the An machine is Ba, the invalid data set is Bb.

2) monitor all disk write operation, the modification of the data on the disk among all pair set Ba is all write in the working area and (is used for storing some data generally at hundreds of million to several G temporarily), supposes with set C and represents the working area.Can safeguard the mapping relations of Ba by certain algorithm to C.Writing of the disk zone of pair set Bb representative can directly be write.

3) carrying out the increment copy will gather the part that newly is used among the part that is used among the C and the set B b exactly and send to An other n-1 platform machines in addition.

4) carry out and to revise IP automatically, current retouching operation is backed up the part that the modification of Ba causes Ba to change.So that variant duplication next time can make all receivers all revert to last state when carrying out the increment copy.Just the 3rd step will at first make the reduction earlier of all receivers.

5) execution in step 2 is to step 5 after.

The implementation method in the 4th step can change mapping into.Use the way of 2 reduction to solve.Our this as can be seen technology takes the form of increment in set C and set B b.And being set B a, the basis of its realization can not be modified.Based on this point, claim that generally this technology is the increment copy.His obvious defects is can not revise Ba, open mode that the software of realizing on this principle can not the supportive protection card and permanent write operation.It takes 2 comparison complicated algorithm such as reduction to realize needs in order to realize Ba and not to be modified but realize revising automatically the operation that IP etc. needs to revise Ba again.Because it does not support open mode to make some need repeat to restart machine, need every machine to have the demand of independent personalized configuration to satisfy with bad.The user also has been subjected to very big restriction to the use of machine.

By above-mentioned introduction as can be known, the thought of increment copy is the write operation of shining upon the data of having transmitted in the disk, copies the data that increase after the data of having transmitted relatively to receiver.The greatest drawback of the realization of increment copy be its in order effectively to monitor increment, must limit the write operation of the data that the last time has been transmitted, otherwise he has not just had the basis of increment.There is the following problem in this implementation method:

1) not house steward's pattern of supportive protection card and permanent backup (because under house steward's pattern, can write arbitrarily) to disk;

2) it must carry out primary network and copies the basis of setting up increment, in case and its lost this basic increment and can not realize;

3) dumb in management, the user must guarantee network-copy was carried out in all machine unifications, has so in a single day gone bad a machine or newly-increased machine, and the increment copy will be done a lot of extra works just can make this machine also support the increment copy;

4) the implementation more complicated of increment copy, the effect of bringing is not ideal enough.

Summary of the invention

The objective of the invention is to copy existing defective at network-copy in the computer system maintenance and increment, a kind of implementation method of variant duplication is provided, makes a system and support still to support permanent write operation and when revising IP automatically, do not need to set up two complicated operations such as restoration point behind the increment copy.

Technical scheme of the present invention is as follows: a kind of implementation method of variant duplication comprises the steps:

(1) determines arbitrarily that in network environment a machine An who has configured system and software is as transmitter, with other n-1 platform machines as receiver, set variable set Dn for copying the data of other machines to, gather at the valid data that erection stage variable set Dn equals in the An machine from a machine;

(2) carry out network-copy for the first time, Dn copies all receivers to the set of the variable in the transmitter An system;

(3) content with the set of the variable in all machines in network environment Dn empties, and does not promptly need the variable data that copies;

(4) data recordng of write operation will take place in the variable set Dn of corresponding machine to the write operation of disk in all machines in the monitor network environment;

(5) carrying out network-copy again, is transmitter with any machine Ai, sends order request for earlier all receivers its variable set Dn is sent to transmitter, and n is not equal to i, and transmitter is asked the variable set of all machines and obtain variable set D;

(6) transmitter Ai obtains the valid data set B a of own disk again according to file system, and the common factor of variable being gathered D and valid data set B a is stored among its variable set Di;

(7) all data among the variable set Di of transmitter Ai are sent to all receivers;

(8) repeating step (3)～(7).

The method of variant duplication provided by the present invention is that the disk that the disk write operation of monitoring room inner machine causes changes, when network-copy, collect all receivers and transmitter then the zone that changes has taken place, these data that zone of variation has taken place are sent to receiving terminal from transmitting terminal.Variant duplication is compared with the increment copy, and change has taken place its basic principle.Variant duplication is only monitored the variation that disk takes place, any restriction is not done in write operation to disk, its basic thought is to remove to recover receiver with the data of transmitter the part that changes has taken place, thereby, make copy procedure more efficient and convenient, compare with other copy mode and have significant superiority.

Description of drawings

Fig. 1 is the flow chart of variant duplication implementation method.

Embodiment

The present invention is described in further detail below in conjunction with drawings and Examples.

The flow chart of the method for the invention as shown in Figure 1.

Suppose and have only a subregion (C) on the hard disk, make this subregion support variant duplication.The step of its realization is as follows:

(1), installation steps

The C dish is divided into set of blocks A by the piece in 64 sectors.Set up variable bitmap (effect of variable bitmap is whether corresponding of expression has new variable data need send to other receiver) simultaneously, in the variable bitmap and the position of piece correspondence is the no variable of 0 expression, is that 1 expression has variable.All positions of variable bit map area are labeled as 1.

(2), to tell which piece in the set A according to the file system scanning area be piece with valid data, which is the piece that does not have valid data.The valid data piece is labeled as set B, non-valid data piece is labeled as set C.Then exist and concern A=B+C.Position corresponding in all corresponding variable bit map area among the set C is labeled as 0.

(3), carry out network-copy for the first time, in the variable bit map area promising 1 piece send to all receivers.

(4), the variable bitmap that upgrades all receivers is 0 (not sending corresponding piece when making next time network-copy).

(5), the user enters operating system with house steward's pattern and carries out normal running.Monitor all disk write operation, the position in the variable bitmap of the piece correspondence that system is write is labeled as 1.

(6), the user carries out permanent write operation.The user is labeled as 1 to the variable bitmap of the piece of the write operation correspondence of disk.

(7), the user carries out network-copy again.Then at first obtain the variable bitmap of receiver from all receivers.The position that is labeled as in the variable bitmap with the receiver of all acquisitions in the variable bitmap of 1 piece corresponding sending terminal is labeled as 1.

(8), be after 1 piece sends to all receivers, all variablees protection bitmaps that participate in the machine of network-copy to be re-set as 0 in the variable bitmap with all transmitting terminals.

(9), operate repeating step (5)～(8) later on.

Claims

1. the implementation method of a variant duplication comprises the steps:

(5) carrying out network-copy again, is transmitter with any machine Ai, sends order request for earlier all receivers its variable set Dn is sent to transmitter, and wherein n is not equal to i, and transmitter is asked the variable set of all machines and obtain variable set D;

(8) repeating step (3)～(7).