Plaque It!
Sponsored by: Flash of Genius |
The present invention relates to a system and method for sharing a storage device among controllers. More specifically, the present invention relates to a system and method where the data sharing process of storage partitions are sped up and potential conflicts between accesses are avoided.
Modern server machines are capable of having multiple servers installed in them and the servers may all couple to a disk array. A disk array consists of several partitions with individual logical unit numbers (LUN) for the servers to perform data access with. The presence of the partitions is detectable to all servers so that the servers can share the data stored in the disk array. No priorities for access are set to avoid errors when the servers perform data access. Thus when any two servers try to access the same partition at the same time, the partition can not determine the priority between those servers, causing conflict in the partition and rendering accessed data invalid or lost.
To prevent the above-mentioned conflict situation while allowing accessibility to any one of the partitions in the disk array, the servers are coupled through an Ethernet network. At least one partition is assigned to a server and the server can only access its own partition(s). If a data access with a partition other than the ones assigned to the server, then a request must be made to the server that owns the partition to perform such data access. For example, when a first server tries to access data in a partition assigned to a second server, the first sever must convert accessing information (including location of the first server, IP address, path to the partition, access link and check sum) to multiple packets in the Ethernet protocol and transmit the packets to the second server. After receiving all the packets, the second server disassembles and analyzes those packets to acquire the information of the target partition and perform data access with that partition accordingly. The accessed data is then converted into multiple packets in the Ethernet protocol and transmitted back to the first server.
However, to share the data in said partition, not only does the communication between the two servers require conversions of information into packets complying with the network protocol, it also relies on the Ethernet network as the media. This type of design increases the complexity among the connections of the servers and the possibility of errors occurring during data access with non-assigned partitions. Furthermore, program developers for such a server machine must first set up the communication protocol. This introduces a higher cost, more time and effort in product development and could ultimately lead to a loss of its competitive edge on the market. Consequently, it is imperative for a server machine manufactures to have a system and method for sharing storage device among controllers that speeds up the sharing of the data in the storage device while sparing the complicated communication procedure.
After considerable research and experimentation, a system and method for sharing storage device among controllers according to the present invention has been developed so as to overcome the drawbacks associated with said prior method.
It is an object of the present invention to provide a system and method for sharing a storage device among controllers, said system comprising a first controller and a second controller, said first and second controllers being coupled together and to a storage device having a plurality of storage units, said first and second controllers being capable of detecting the presence of said plurality of storage units via said storage device and defining the storage units detected by the other controller as virtual storage units, and via the other controller and according to identification numbers of said storage units, said first and second controllers defining said virtual storage units with unique virtual identification numbers, said first and second controllers each further comprising an allocating unit for specifying storage units accessible by the corresponding controller. When said first controller is trying to access a storage unit not directly accessible by said first controller, said first controller converts the identification number of said storage unit into a virtual identification number, and requests said second controller to perform data access with said virtual storage units represented by said virtual identification number. As a result, data sharing process of said storage partitions are sped up and potential conflicts between accesses are avoided.
It is another object of the present invention that said first and second controllers each further comprise a first port and a second port, said first port of said first controller and said first port of said second controller are coupled together, said storage device further comprises a third port, said third port is coupled to said second port of said first controller and said second port of said second controller. When said first and second controllers are activated, via said storage device, said controllers obtain the access path of said storage units according to each of said identification numbers, then through said first ports, detect the storage units already detected by the other controller. According to the detected identification numbers of said storage units, the other controller determines whether said detected storage units have duplicate identification numbers. Said controllers then use a base number and said identification numbers in calculating said virtual identification numbers of each of said virtual storage units to obtain access paths to each of said storage units via said virtual identification numbers.
The above and other objects, features and advantages of the present invention will become apparent from the following detailed description taken with the accompanying drawings.
FIG. 1 shows a functional block diagram of a system according to the present invention;
FIG. 2 shows the flow chart of the steps taken by a first controller when accessing data in any one of the storage units in a system according to the present invention;
FIG. 3 shows the flow chart of the steps taken by a second controller when a request signal is received in a system according to the present invention;
FIG. 4 shows the flow chart of the steps taken by the controllers when said controllers are activated and storage units are detected in a system according to the present invention;
FIG. 5 shows a block diagram of an access indication table in a system according to the present invention;
FIG. 6 shows the detailed flow chart of the step 404 in FIG. 4.
FIG. 7 shows a diagram of a path mapping table a system according to the present invention;
FIG. 8 shows the flow chart of FIG. 7.
FIG. 9 shows the flow chart of the steps taken by a first controller when it determines the target storage unit is not directly accessible in a preferred embodiment of the present invention;
FIG. 10 shows the flow chart of the steps taken by a second controller when a request signal is received in the embodiment of FIG. 9.
Serial Attached SCSI (SAS) is the successor of the Small Computer System Interface (SCSI). It has gradually replaced SCSI as the next standard in storage interface. Comparing to parallel SCSI, transmitting data in serial improves the efficient, expandability and utilization. The biggest advantage is that SAS system architecture accommodates the integration of 3.5-inch Atlas SAS hard drives spinning at 10,000 or 15,000 RPM and Serial ATA hard drives. SAS storage devices can be used as an universal building block in constructing the basic storage structure of a data center to fulfill the requirements of on-line data. SAS has a maximum data transfer rate of 3.0 Gb/sec to achieve performance at a higher level, and a hard drive can be connected at the second port to upgrade the utilization and reliability of the overall system.
Refer to FIG. 1 for a system for sharing storage device among controllers according to the present invention. The system comprises at least a first controller 1 (1 st SAS Controller) and a second controller 10 (2 nd SAS controller). Said first and second controllers 1 and 10 are coupled and communicate to each other through a first SAS channel 13 . Said first and second controllers are further coupled to a storage device 2 (e.g. a disk array) through a second SAS channel 14 . Said storage device unit 2 has a plurality of storage units 20 , each represented by a Logical Unit Number (LUN). Via said second SAS channel 14 , said first and second controllers 1 and 10 are capable of detecting the presence of said storage units 20 in said storage device 2 . Said first and second controllers 1 and 10 each can also detect via the other controller the storage units detected by the other controller. To avoid any conflicts and confusion between storage units 20 detected by itself and the storage units 20 detected by the other controller, the storage units detected by the other controller are defined as a plurality of virtual storage units.
Refer to FIG. 1. Said first controller 1 further comprises a first allocating unit 3 and said second controller 10 further comprises a second allocating unit 4 for specifying storage units 20 directly accessible by their corresponding controller. When said first controller 1 is trying to access a storage unit 20 directly accessible by said second controller 20 , said first controller 1 requests said second controller 10 to perform data access with said virtual storage units. Said second controller 10 then finds the storage unit 20 which is presented as said virtual storage unit according to said resource allocating unit 4 , proceeds with data access with said storage unit 20 and transmits the result of said data access back to said first controller 1 .
Refer to FIG. 1 for a preferred embodiment according to the present invention. Said first and second controllers 1 and 10 each further comprises a first SAS port 11 and a second SAS port 12 . Said first controller 1 and said second controller 10 are coupled with their respective first SAS ports to form a first SAS channel 13 , and said two controllers are both further coupled to a third SAS port 21 of said storage device 2 with their respective second SAS ports to form a second SAS channel 14 as the bridge of communication between said two controllers.
Each storage unit 20 has an unique identification number 200 . Refer to FIG. 1 again. According to said identification numbers 200 and via said second SAS channel 14 , said first and second controllers 1 and 10 are capable of detecting the presence of said storage units 20 in said storage device 2 . With the information of said identification numbers 200 , said controllers then each defines the virtual storage units with a set of virtual identification numbers that are different from said identification numbers 200 as a reference to the storage units detected by the other controller.
Refer to FIG. 2 for the flow chart of the steps taken by said first controller 1 when trying to access data in any one of the storage units 20 .
Refer to FIG. 3 for the flow chart of the steps taken by said second controller 10 when a request signal is received.
Refer to FIG. 4 for the flow chart of the steps taken by the controllers 1 and 10 after said controllers are activated to obtain the access path of said storage unit 20 according to the identification number 200 of each of said storage units 20 .
Consequently, said first or second controller 1 or 10 may have two sets of storage device 2 while still be able to correctly distinguish between the storage units 20 of its storage device 2 and the storage units 20 of the other controller 10 or 1 , effectively avoiding potential conflicts between accesses to the same storage unit.
Refer to FIGS. 1 and 5 again. In said preferred embodiment, said resource allocating units 3 and 4 can be an access indication table 30 comprising an identification number field 31 and an authority field 32 . Said identification number field 31 records identification numbers 200 of said storage units 20 in order and said authority field 32 has a flag 33 to indicate according to each identification number 200 whether a direct access with a storage unit 20 is allowed. As depicted in FIG. 6, when said first controller 1 tries to access data in one of said storage units 20 and determines whether a direct data access with said storage unit 20 is allowed according to said indication table 30 , the following steps are performed.
Refer to FIGS. 1 and 7 for another preferred embodiment. Said resource allocating units 3 and 4 is a path mapping table 40 comprising an identification number field 41 for recording identification numbers 200 of said storage units in order and a path field 42 for specifying an access path (e.g. storage unit # 1 ) to a storage unit 20 corresponding to said first and second controller 1 and 10 . Storage units 20 not associated with said first controller 1 is specified with an access path corresponding to said second controller 10 in their path fields and vise versa. When said first controller 1 tries to access data in any one of said storage units 20 and determines whether direct data access with said storage unit 20 is allowed according to said path mapping table 40 , said first controller 1 performs the steps shown in FIG. 8.
Refer to FIGS. 1 and 9. When said first controller 1 determines a direct data access with said storage unit 20 is not allowed, said first controller 1 performs the following steps.
When said second controller 10 receives said request signal, said second controller 10 performs the following steps.
Refer to FIG. 1 again. The maximum capacity of said storage device 2 is 128 storage units 20 (the base number is 128). Only four storage units 20 are installed in said storage device 2 and the identification numbers 200 of said four storage units 20 are 0 , 1 , 2 and 3 . Via said storage device 2 and said second controller 10 , said first controller 1 detects said storage units 0 , 1 , 2 , 3 and the access path of virtual storage units 0 , 1 , 2 and 3 . Having storage units 20 with the same identification number 200 may potentially cause conflicts between said first and second controllers 1 and 10 . Therefore, said first controller 1 treats the storage unit 20 with the same identification number 200 as a virtual storage unit and starts the identification number 200 from 128 and up, assigning the virtual identification numbers of said virtual storage units as 128 , 129 , 130 and 131 in order. With this scheme, said first and second controller 1 and 10 each will have access paths to the storage units 20 as 0 , 1 , 2 , 3 , 128 , 129 , 130 , and 131 .
Refer to FIGS. 1 and 7 again. Said identification number field 41 records 0 , 1 , 2 and 3 in order, with 0 and 1 indicating the represented storage units 20 are defined to be directly accessible by said first controller 1 while 2 and 3 indicating the represented storage units 20 are defined to be directly accessible by said second controller 10 . Columns 0 and 1 of said path field 42 in said first controller 1 record the access paths to storage units number 0 and 1 , respectively, and columns 2 and 3 of said path field 42 in said first controller 1 record the access paths to said second controller 10 . On the other hand, columns 0 and 1 of said path field 42 in said second controller 10 record the access paths to said first controller 1 , and columns 2 and 3 of said path field 42 in said second controller 10 record and the access paths to storage units number 2 and 3 , respectively.
Refer to FIG. 1 again. As an example, when said first controller 1 receives an external signal to request data access to the storage unit in column 3 (storage unit number 3 ), said first controller 1 first reads the path field 42 of its first resource allocating unit 3 and determines said first controller 1 is not allowed to access said storage unit 20 directly. Said first controller 1 then converts the identification of storage unit number 3 into storage unit number 131 and requests said second controller to perform data access with storage unit number 131 . Said second controller 10 converts storage unit number 131 back to storage unit number 3 and according to the path field 42 in said second resource allocating unit 4 , determines direct access with said storage unit 20 is allowed. Hence said second controller 10 performs data access with said storage unit 20 and returns the data accessed back to said first controller 1 .
While the invention herein disclosed has been described by means of specific embodiments, numerous modifications and variations could be made thereto by those skilled in the art without departing from the scope and spirit of the invention set forth in the claims.