Embodiment
Hereby with regard to a preferred embodiment of the present invention, coordinate graphicly, be described in detail as follows.
The invention provides a kind of high in the clouds cluster system (cluster system, will be in instructions in literary composition referred to as this system) below, and the start dispositions method that uses of this high in the clouds cluster system.By the present invention, in a blank main frame is added into this system time, can start shooting rapidly, and be deployed the role required into this system, for example manage end points (management node), computing end points (compute node), store end points (storage node) or web service endpoint (website node) etc., be not limited.And, in the time that managerial personnel's wish changes wherein the role of a main frame, only need re-start deployment to this main frame, can directly change the role of this main frame, this main frame does not need to reinstall the required operating system of new role and application program.Thus, can effectively shorten main frame and add the time in this system to.
Head refers to Fig. 1, is the system architecture diagram of a preferred embodiment of the present invention.This system of the present invention mainly comprises a start server (boot server) 1, a management server (management server) 2, a system storage pool (system storage pool) 3 and at least one main frame 4.It is worth mentioning that, before this main frame 4 is not yet added into this system, be a completely blank main frame.
Please refer to Fig. 2, it is the rack schematic diagram of the high in the clouds machine room of a preferred embodiment of the present invention.As shown in the figure, this start server 1, this management server 2, this system storage pool 3 and this main frame 4 are mainly arranged among the same rack 60 of high in the clouds machine room, connect each other by the network switch (Switch) 6 in this rack 60.But, above are only a preferred embodiments of the present invention, this start server 1, this management server 2, this system storage pool 3 and this main frame 4 also can be arranged at respectively different racks or different data centers, and are connected to each other by network each other, are not limited.Moreover, this system storage pool 3 is assembled by the multiple storage areas in this system (as the hard disk in multiple servers) the large-scale storage area forming, in the present embodiment graphic, present in the mode of single station server, to facilitate explanation, but this system storage pool 3 is not limited to the server of single.
Please refer to Fig. 3, it is the system block diagrams of a preferred embodiment of the present invention.This start server 1 is after this main frame 4 is unlocked, the very first time provides this main frame 4 one network start rules (netboot policy), thus, this main frame 4 can be carried out a network boot program according to this network start rule, the for example Pre-boot Execution Environment of a standard (Preboot eXecution Environment, PXE) program, but be not limited.
This start server 1 also can be by the connectivity port on it (figure indicate), for example USB (universal serial bus) (Universal Serial Bus, USB) connectivity port or external series hi-tech configuration (external Serial Advanced Technology Attachment, e-SATA) connectivity port etc., the hard disk 5 of pegging graft.More specifically, this hard disk can be a portable memory device that possesses hot plug (Plug and Play) function.In this hard disk 5, store at least one image file (Golden image) 51, this image file 51 comprises complete operating system, and the required content of all roles in this system, as application program (applications), file (files), default (system configuration) and system application (system utilities) etc., be not limited.
When this main frame 4 is carried out this network boot program to particular step, can download and directly use this image file 51 by this start server 1, to complete whole network boot program.In other words, this system, for the role that will serve as after 4 starts of this main frame, is not carried out corresponding installation action, but content required all roles is all contained in this image file 51, and offers this main frame 4 and use.Thus, this main frame 4 needn't be installed complete operating system in addition, therefore can effectively save the on time of this main frame 4.
This system storage pool 3, by the multiple storage areas in this system are formed, in order to offer this main frame 4 store data, after these main frame 4 unlatchings, can be connected to this system storage pool 3 to use the space of this system storage pool 3.And, in this system storage pool 3, there is at least one root directory file system (root file system) 301, when this main frame 4 is connected to after this system storage pool 3, also can obtain and use this root directory file system 301.Thus, all data relevant to this system that this main frame 4 is performed, all can be stored in this system storage pool 3, and needn't be stored in the hard disk (figure does not indicate) of this main frame 4 oneself.More specifically, this main frame 4 mainly can connect this system storage pool 3 by the mode of network carry (mount), this system storage pool 3 is considered as to a local side hard disk of this main frame 4, and uses the space of this system storage pool 3.
More specifically, this system storage pool 3 mainly can be separated out multiple file 30, and respectively this file 30 offers respectively this different main frames 4 and uses.This main frame 4 mainly can be connected to one of them of the plurality of file 30 in this system storage pool 3, to obtain this root directory file system 301 in this file 30, and uses the space of this file 30.Thus, this system storage pool 3 can offer multiple host 4 simultaneously and use, and respectively the data of this main frame 4 can be not interfering with each other, do not have the problem that data leak.
Use the benefit of these system storage pool 3 maximums to be, the data that this main frame 4 is relevant to system all exist in this corresponding data folder 30, therefore in the time that this main frame 4 is damaged, as long as add a backup host, allow this backup host carry out this identical network boot program, be connected to same this data folder 30 in this system storage pool 3, finally download again and use this image file 51, can reduce easily this main frame 4.
It is worth mentioning that, same this image file 51 is all downloaded and used to the All hosts 4 in this system, and therefore main frame 4 completes after start, will have duplicate content, does not have role's difference.Be with, complete after start at this main frame 4, need further be connected to this management server 2 to register, thus, this main frame 4 can be accepted the deployment action (deployment) that this management server 2 carries out, and has made this main frame 4 of deployment can directly serve as the corresponding role in this system.
As shown in the figure, this management server 2 has a control interface 20, this control interface 20 mainly can be one take network as basis (web-based) user interface (User Interface, or command interface (Command Line Interface UI), CLI), operate for managerial personnel.Managerial personnel can pass through the operation to this control interface 20, this main frame 4 is carried out to a required role's setting, thus, this control interface 20 can be according to managerial personnel's setting content, produce a corresponding configuration file (profile) 302, wherein this configuration file 302 has recorded the required content of this role.
After managerial personnel set, this management server 2 starts a deployment program immediately,, this management server 2 can enter in this main frame 4 automatically, and according to this configuration file 302, in this image file 51 that this main frame 4 of activation (enable) uses, correspond to this role's content.Thus, this main frame 4 is accepting, after the deployment of this management server 2, can to become this role that these managerial personnel set.
As mentioned before, in this image file 51 that this main frame 4 is downloaded and used, comprise in fact the required content of all roles in this system, the deployment action of this management server 2, mainly to see this main frame 4 is set as what role by managerial personnel, be this role's of activation corresponding content, the unwanted content of this role can be by decapacitation (disable).But, in this main frame 4, still remaining with the required content of other roles, these can't be deleted by the content of decapacitation.That is to say, in the time that managerial personnel will change the role of this main frame 4, as long as reset, this control interface 20 will produce this new configuration file 302, this management server 2 is in the time disposing action, will be according to new content corresponding to these configuration file 302 activations, and other content of decapacitation.Therefore, this main frame 4 can be redeployed the role for other easily.
It is worth mentioning that, for can be in the time that this main frame 4 be damaged this main frame 4 of fast restore, therefore, after this configuration file 302 produces, this management server 2 can be stored to this configuration file 302 in the corresponding folder 30 in this system storage pool 3 that this main frame 4 connects use.Thus, when a backup host starts and is connected to after same this file 30, this management server 2 this configuration file 302 in can this file 30 of direct basis is disposed, and does not need managerial personnel to re-start setting.
This main frame 4 is after these management server 2 registrations, this management server 2 can learn that this main frame 4 has completed start, therefore, this system can be monitored by this management server 2 operating state of this main frame 4, such as flow (traffic), handling capacity (throughput), CPU usage, temperature or humidity etc.This management server 2 can produce a status information 303 according to the operating state of this main frame 4, and regularly this status information 303 is stored in this file 30 in this system storage pool 3 that this main frame 4 connects use.Moreover this management server 2 also can pass through this control interface 20, shows above-mentioned this configuration file 302 and this status information 303 with word or graphics mode.Thus, managerial personnel can check the every terms of information of this main frame 4 by this control interface 20, and then are of value to the management for whole system.
The continuous Fig. 4 that refers to, for the image file of a preferred embodiment of the present invention is made process flow diagram.First, managerial personnel can, by extra server (figure does not indicate), install the required operating system of this system, and the required various contents of various role, and then, make this image file 51 (step S10) by these operating systems and content.And, after this image file 51 completes, this image file 51 is stored in this hard disk 5 (step S12).Then, judge whether also to need to make other image file (step S14), if so, re-execute this step S10 ~ this step S12, to make and to store other image file.If image file completes, this hard disk 5 is plugged on this start server 1 (step S16).
The object of above-mentioned steps is, must not only include a this image file 51 in this hard disk 5, also may comprise this multiple image files 51.This main frame 4 can be downloaded this multiple image files 51 simultaneously, and again according to the demand of this system, uses this corresponding image file 51.But the foregoing is only preferred embodiments of the present invention, should be as limit.
Consulting Fig. 5, is the start process flow diagram of a preferred embodiment of the present invention.In the time that this main frame 4 of a blank is added in this system, (for example insert the slot in this rack 60), can connect this start server 1, this management server 2 and this system storage pool 3 by this network switch 6.After this main frame 4 adds, mainly need first open this main frame 4 (step S20), wherein, managerial personnel can press the entity starting key (figure does not indicate) on this main frame 4, or by online start (Wake-On-LAN, WOL) mode, opens in this main frame 4 of far-end control, is not limited.
After this main frame 4 is unlocked, first obtain a spendable IP address and a Hostname (step S22), then use this IP address to be connected to this start server 1, and obtain this network start rule (step S24) by this start server 1, thus, this main frame 4 just can, according to this IP address, this Hostname and this network start rule, be carried out this network boot program.Then, this main frame 4 can connect this to system storage pool 3, and request is connected to this system storage pool 3 (step S26), to obtain this root directory file system 301, and uses the space of this system storage pool 3.More specifically, this main frame 4 is connected to the corresponding folder 30 in this system storage pool 3, to obtain this root directory file system 301 in this file 30, and uses the space of this file 30.
This main frame 4 is obtained after this root directory file system 301, can again be connected to this start server 1, and downloads and use this image file 51 (step S28) in this hard disk 5 by this start server 1, thus, completes this network boot program.Finally, complete after start at this main frame 4, can register (step S30) to this management server 2, to accept the deployment action (step S32) of this management server 2, and then after deployment completes, serve as the required role of this system.
Continuing and refer to Fig. 6, is the deployment process flow diagram of a preferred embodiment of the present invention.As mentioned before, when there being this multiple main frames 4 to complete after start, can have identical content, this is because this image file 51 that main frame 4 uses has comprised the required content of all roles in this system.Therefore,, if will allow this main frame 4 can serve as the specific role in this system, need to do further deployment by this management server 2 and move.First, this management server 2 provides this administration interface 20, thus, is accepted managerial personnel's peripheral operation by this administration interface 20, this main frame 4 is carried out to this required role setting (step S40).
Then, this control interface 20 can produce above-mentioned this configuration file 302 (step S42) according to this setting, and wherein, this configuration file 302 has recorded the required content of this role.And, after setting completes, this management server 2 can start a deployment program (step S44), by this deployment program, this management server 2 can enter in this main frame 4 automatically, and according to this configuration file 302, in this image file 51 that this main frame 4 of activation uses, correspond to this role's content (step S46).Thus, this main frame 4 can complete after deployment, becomes this role (step S48) that managerial personnel set.It is worth mentioning that, this image file 51 also predeterminable (default) by whole all the elements activations, and this configuration file 302 records the unwanted content of this role, this management server 2 is again according to this configuration file 302, in this image file 51 that this main frame 4 is used, do not belong to this role's the whole decapacitation of content.But, the foregoing is only a preferred embodiment of the present invention, should be as limit.
Continuing and refer to Fig. 7, is the system block diagrams of another preferred embodiment of the present invention.In the present embodiment, this system also includes a store pond 7, connects this main frame 4, and more specifically, this network switch 6 that this store pond 7 connects in this rack 60, and connect this start server 1, this management server 2 and this main frame 4 by this network switch 6.The large-scale storage area forming is mainly assembled by the multiple storage areas in this system (as the blank hard disk in multiple host 4) in this store pond 7, in the present embodiment graphic, present in the mode of single station server, to facilitate explanation, but this store pond 7 is not limited to the server of single.
Be with the difference of this system storage pool 3, this system storage pool 3 provides this main frame 4 certain space, in order to deposit the data relevant to system in this main frame 4; This store pond 7 offers this main frame 4 and has more more greatly flexible storage area, to make this main frame 4 be able to generate, use every application program on this store pond 7.
In the present embodiment, this control interface 20 mainly can be used as a solid data center management module (the Physical Data Center Management in this system, PDCM), for example, in order to control and to manage all entity machines (this main frame 4) in this system.And as shown in Figure 7, on this control interface 20, also there is a virtual data center administration module (Virtual Data Center Management, VDCM) 200, in order to manage all virtual machines in this system.
Managerial personnel system can manipulate this control interface 20, and further uses this virtual data center administration module 200 on this control interface 20.This virtual data center administration module 200 can be accepted outside manipulation, to specify this main frame 4 to carry out the required virtual machine 70 of generation system, and, this generated virtual machine 70 is stored in this store pond 7.Wherein, this virtual machine 70 mainly can produce according to this image file 51 of these main frame 4 use, and therefore this system needn't, in order to produce this virtual machine 70, additionally provide the installation procedure of operating system and application program.
By the present invention, managerial personnel can pass through the operation of this virtual data center administration module 200, control easily which platform main frame 4 and will produce this virtual machine 70, produce several these virtual machines 70, and this virtual machine 70 will be served as any role, therefore concerning managerial personnel, the present invention can provide quite powerful convenience in fact.
Please refer to Fig. 8, be the virtual machine product process figure of a preferred embodiment of the present invention.As mentioned above, first, need provide this virtual data center administration module 200 (step S50) by this control interface 20 in this management server 2, then, this virtual data center administration module 200 can be accepted outside manipulation, to specify this main frame 4 to generate this virtual machine 70 (step S52).This virtual machine 70 is to generate this virtual machine 70 according to this image file 51 of these main frame 4 use, and this generated virtual machine 70 is stored in this store pond 7 (step S54).Wherein, this store pond 7 also can be divided into multiple files, and to connect for this different main frames 4 respectively, its connection/carry mode is identical with this system storage pool 3, does not repeat them here.
Consulting Fig. 9, is the system block diagrams of another preferred embodiment of the present invention.As shown in the figure, this system also includes a distribution (assign) server 8, connects this start server 1, this management server 2, this system storage pool 3 and this main frame 4.In the present embodiment, main system is distributed, is managed IP address and the Hostname of the All hosts 4 in this system by this distribution server 8.
In the present embodiment, this main frame 4 is one of them port (figure does not indicate) connecting on this network switch 6, and this distribution server 8 is also connected to this network switch 6, thus, this main frame 4 can connect this distribution server 8 by this network switch 6, to accept the service of this distribution server 8.Managerial personnel can set this management server 2 in advance, and to make the topology data (topology) 21 that store this system in this management server 2, this distribution server 8 can be connected to this management server 2, and reads this topology data 21.Thus, this distribution server 8 can be according to these topology data 21, in advance for each port on this network switch 6 all distributes an IP address and a Hostname.Thus, in the time of one of them port that this main frame 4 connects on these network switchs 6, this IP address and this Hostname that can use these ports be assigned with to these distribution server 8 requests.
It is worth mentioning that, this distribution server 8 mainly can be according to these topology data 21, for each port on each network switch in the machine room of high in the clouds is all set exclusive IP address and a Hostname, in the present embodiment, only illustrate with this network switch 6, but not as limit.
In this step S22 of earlier figures 5, this main frame 4 is connected to this distribution server 8, files a request, and after by request to this distribution server 8, obtains and IP address and Hostname that this port of using this main frame 4 to connect is assigned with.Thus, this main frame 4 can be carried out this network boot program by this IP address and this Hostname.More specifically, this distribution server 8 produces a corresponding table 80 according to all IP address and the Hostname that produce, in the time that this main frame 4 is filed a request to this distribution server 8, this distribution server 8 these correspondence tables 80 of inquiry, this IP address and this Hostname that are assigned with to obtain this port that this main frame 4 connects.
Please refer to Figure 10 and Figure 11, be respectively the allocation flow figure of a preferred embodiment of the present invention and another preferred embodiment.First as shown in figure 10,, before this System Operation, managerial personnel must set these topology data 21 of this system, and are stored in this management server 2.Thus, after this distribution server 8 is reached the standard grade, can obtain this topology data 21 (step S60) by this management server 2, and then can, according to these topology data 21, all distribute an IP address and a Hostname (step S62) to each port on this network switch 6.Finally, for convenient inquiry, this distribution server 8 can produce this correspondence table 80 according to these IP addresses and Hostname, and is stored in this distribution server 8 (step S64).
Then as shown in figure 11, when this main frame 4 is unlocked, and before will carrying out this network boot program, this main frame 4 need be asked an IP address and a Hostname (step S70) to this distribution server 8.This distribution server 8 receives after this request, inquires about this correspondence table 80 (step S72) according to this request.And then this distribution server 8 can be according to Query Result, obtain this IP address and this Hostname that this port that this main frame 4 connects is assigned with, and this IP address and this Hostname are offered to this main frame 4 use (step S74).After this step S74, this main frame 4 just can be carried out this network boot program with this IP address and this Hostname.
Finally referring to Figure 11, is start and the deployment sequential chart of a preferred embodiment of the present invention.As shown in the figure, relate to the assembly (component) of start dispositions method of the present invention, mainly comprised this start server 1, this management server 2, this system storage pool 3, this main frame 4, this store pond 7 and this distribution server 8.
First, this main frame 4 first inserts in arbitrary slot of this rack 60, is connected, and connects the assembly 1,2,3,7,8 in this system by this network switch 6 with this network switch 6.After this main frame 4 is unlocked, first to this distribution server 8 request of sending (step S80).This distribution server 8 receives after this request, this correspondence table 80 is inquired about, and then obtained an IP address and a Hostname that can use to this main frame 4, and reply to this main frame 4 (step S82).This main frame 4 has been obtained after IP address and Hostname, can be to this this network start rule (step S84) of start server 1 request, and this start server 1 receives after this request, reply this network start rule to this main frame 4 (step S86).After this step S86, this main frame 4 can, according to this IP address, this Hostname and this network start rule, be carried out this network boot program.
Then, this main frame 4 is connected to this system storage pool 3, request connects this system storage pool 3 (step S88) of use, this system storage pool 3 receives after this request, be this main frame 4 and distribute this exclusive file 30, and an index (pointer) of replying these file 30 positions of indication is to this main frame 4 (step S90).Thus, this main frame 4 can be according to this index, is connected in this system storage pool 3, is specific to this file 30 that this main frame 4 uses, and obtains this root directory file system 301 in this file 30, and uses the space of this file 30.Obtain after this root directory file system 301, this main frame 4 can be connected to this start server 1 again, this image file 51 (step S92) is downloaded in request, and after this start server 1 is agreed to, download and use this image file 51, and then complete whole this network boot program (step S94).
When this main frame 4 completes after start, can register (step S96) to this management server 2, so as to, accept the monitoring of this management server 2, and accept the deployment action (step S98) that this management server 2 carries out.The role that this management server 2 is set according to managerial personnel disposes this main frame 4, therefore after this step S98, in this image file 51 that this main frame 4 uses, the content that corresponds to this role can be enabled, and then this main frame 4 can become this role that managerial personnel set after deployment.Finally, in the time that managerial personnel have demand, this main frame 4 also can connect and uses this store pond 7, and according to the appointment of this virtual data center administration module 200, produces this virtual machine 70, and be stored in this store pond 7.
By the present invention, this main frame 4 is added into after this system, only need to download this image file 51 once, in the time that managerial personnel will change the role of this main frame 4, only need carry out by this management server 2 content of this image file 51 that this main frame 4 of activation/decapacitation uses, can change easily the role of this main frame 4, this main frame 4 needn't be downloaded this image file 51 again, and also needn't carry out the installation action of extra operating system or application program.Therefore, can make the construction speed of whole high in the clouds cluster system more quick, and more simple.
In addition, the present invention makes this main frame 4 connect and uses this system storage pool 3, and all data relevant to system are all stored in this system storage pool 3, therefore, in the time that this main frame 4 is damaged, only need to look for a backup host to be connected to same this file 30 in this system storage pool 3, can this main frame 4 of fast restore, quite facility.Moreover, connection by this store pond 7 is used, and the operation of this virtual data center administration module 200, and managerial personnel can specify this main frame 4 to produce, delete, move this virtual machine 70 easily, for whole high in the clouds cluster system, there is in fact greatest help.
The foregoing is only preferred embodiments of the present invention, non-ly therefore limit to the scope of the claims of the present invention, therefore such as use the equivalence that content of the present invention is done to change, be all in like manner all contained in scope of the present invention, close and give Chen Ming.