Prouduct Center Prouduct Center
Business continuity your location：Home > Product center > Storage System > Storage software > Business continuity >
Product introduction
Technical parameters
Download product information
Dual active controller mode
The valley storage system is designed in a dual active controller (Active/Active) mode. Dual active controller mode enables two controllers to work in parallel, improving system performance, and achieving failover (Failover) to enhance system safety and reliability.。
Failover
The dual controller storage system performs the state detection of the other controller every 10ms. If it exceeds a certain time (generally set to 5 10ms, the specific time interval is determined according to different business applications), the response of the other controller is not detected, then it is considered The other controller fails and the Failover program is started. Through the ICL, the data in the fault controller write cache is mirrored to the local Cache, so that the dual-machine write cache data is consistent, and the user operation being performed by the fault controller is taken over to implement the Failover. The D-JET 6000 dual controller storage system completes failover in less than 8 seconds. If the faulty controller is detected to resume normal operation, the takenover load is transferred, and the dual controller is concurrently operated.。
dynamic data path
In a dual controller system, each logical disk (LD, see section 4.4)“Storage resource virtualization”）Both are connected to two controllers, but only one controller has priority control over the LD. This controller is called the Host Controller and the other controller is the Slave Controller. The host controller controls the Cache of the LD and controls the access channel of the physical disk contained in the LD. When the slave controller receives an I/O request from the client for the LD, the request is handed over to the LDs host controller for processing. In the process of the I/O request to the LD, the client refers to the link between the host controller and the client as the priority access path (Prior Access Link).）。
During the client data read process, the transfer path of the read data will be transmitted on the request path. In the process of client data writing, if the client writes from the priority access path, it needs to perform write cache mirroring to maintain dual controller write cache consistency. If the client writes from a non-priority path, the write data is transmitted on the request path, and the write cache mirror is implemented during the transfer.。
Each logical disk has an attribute to identify which controller is currently the host controller. This flag is dynamically changed during system operation, such as during Failover and Failback. The clients device driver should be designed to accommodate this changing I/O driver so that the client sends most of the I/O requests over the priority access link to the LDs current host controller, reducing ICL in heavy load environments. Frequent communication affects performance issues. Very good performance improvement. The use of dynamic data paths can achieve parallel operation of two controllers, improve system performance, and simplify data transmission paths, reduce complexity, and enhance data integrity.Each logical disk has an attribute to identify which controller is currently the host controller. This flag is dynamically changed during system operation, such as during Failover and Failback. The clients device driver should be designed to accommodate this changing I/O driver so that the client sends most of the I/O requests over the priority access link to the LDs current host controller, reducing ICL in heavy load environments. Frequent communication affects performance issues. Very good performance improvement. The use of dynamic data paths can achieve parallel operation of two controllers, improve system performance, and simplify data transmission paths, reduce complexity, and enhance data integrity.。
With battery write Cache and write cache image
The valley storage system can be configured with 32~512GB ECC read cache and 32GB with battery write cache (Write Back mode). In the process of data writing, if the power failure occurs, the battery write cache can ensure that the data in the Cache is not lost for a period of time (usually 72 hours), and the related data can still be used normally after the system is started. System data loss caused by power failure, maximizing system data integrity。
In the process of data writing, the write cache can also effectively improve the operation of the back-end hard disk. Small blocks of data that are written multiple times can be integrated into one large block of data, reducing the number of hard disk reads and writes. For example, multiple operations on a RAID can be combined into one full strip write, so that it is not necessary to read the old data required for striping from the hard disk, and improve the I/O operation efficiency.。
A dual controller system deploys a write cache image between controllers. When the host writes the cache write data to one of the controllers, the written data is automatically transferred to the other controller write cache through the 80Gb/s ICL. After the mirroring is successful, the system will confirm the write success to the host. . This mechanism ensures the consistency of the dual controller data during the data writing process, preventing data loss and business interruption due to single controller damage.。
In a dual controller system, if one controller is in the process of data writing, such as a power failure, another controller starts the Failover handler, mirroring the data in the write buffer of the fault controller to the local write cache to prevent Data loss, taking over the business that the faulty controller is running, does not affect the normal use of the user.。
The storage administrator can turn off the image caching mechanism as needed to further improve system performance. Of course, this will result in a decrease in system security and business continuity.。
The valley storage system is designed in a dual active controller (Active/Active) mode. Dual active controller mode enables two controllers to work in parallel, improving system performance, and achieving failover (Failover) to enhance system safety and reliability.。
Failover
The dual controller storage system performs the state detection of the other controller every 10ms. If it exceeds a certain time (generally set to 5 10ms, the specific time interval is determined according to different business applications), the response of the other controller is not detected, then it is considered The other controller fails and the Failover program is started. Through the ICL, the data in the fault controller write cache is mirrored to the local Cache, so that the dual-machine write cache data is consistent, and the user operation being performed by the fault controller is taken over to implement the Failover. The D-JET 6000 dual controller storage system completes failover in less than 8 seconds. If the faulty controller is detected to resume normal operation, the takenover load is transferred, and the dual controller is concurrently operated.。
dynamic data path
In a dual controller system, each logical disk (LD, see section 4.4)“Storage resource virtualization”）Both are connected to two controllers, but only one controller has priority control over the LD. This controller is called the Host Controller and the other controller is the Slave Controller. The host controller controls the Cache of the LD and controls the access channel of the physical disk contained in the LD. When the slave controller receives an I/O request from the client for the LD, the request is handed over to the LDs host controller for processing. In the process of the I/O request to the LD, the client refers to the link between the host controller and the client as the priority access path (Prior Access Link).）。
During the client data read process, the transfer path of the read data will be transmitted on the request path. In the process of client data writing, if the client writes from the priority access path, it needs to perform write cache mirroring to maintain dual controller write cache consistency. If the client writes from a non-priority path, the write data is transmitted on the request path, and the write cache mirror is implemented during the transfer.。
Each logical disk has an attribute to identify which controller is currently the host controller. This flag is dynamically changed during system operation, such as during Failover and Failback. The clients device driver should be designed to accommodate this changing I/O driver so that the client sends most of the I/O requests over the priority access link to the LDs current host controller, reducing ICL in heavy load environments. Frequent communication affects performance issues. Very good performance improvement. The use of dynamic data paths can achieve parallel operation of two controllers, improve system performance, and simplify data transmission paths, reduce complexity, and enhance data integrity.Each logical disk has an attribute to identify which controller is currently the host controller. This flag is dynamically changed during system operation, such as during Failover and Failback. The clients device driver should be designed to accommodate this changing I/O driver so that the client sends most of the I/O requests over the priority access link to the LDs current host controller, reducing ICL in heavy load environments. Frequent communication affects performance issues. Very good performance improvement. The use of dynamic data paths can achieve parallel operation of two controllers, improve system performance, and simplify data transmission paths, reduce complexity, and enhance data integrity.。
With battery write Cache and write cache image
The valley storage system can be configured with 32~512GB ECC read cache and 32GB with battery write cache (Write Back mode). In the process of data writing, if the power failure occurs, the battery write cache can ensure that the data in the Cache is not lost for a period of time (usually 72 hours), and the related data can still be used normally after the system is started. System data loss caused by power failure, maximizing system data integrity。
In the process of data writing, the write cache can also effectively improve the operation of the back-end hard disk. Small blocks of data that are written multiple times can be integrated into one large block of data, reducing the number of hard disk reads and writes. For example, multiple operations on a RAID can be combined into one full strip write, so that it is not necessary to read the old data required for striping from the hard disk, and improve the I/O operation efficiency.。
A dual controller system deploys a write cache image between controllers. When the host writes the cache write data to one of the controllers, the written data is automatically transferred to the other controller write cache through the 80Gb/s ICL. After the mirroring is successful, the system will confirm the write success to the host. . This mechanism ensures the consistency of the dual controller data during the data writing process, preventing data loss and business interruption due to single controller damage.。
In a dual controller system, if one controller is in the process of data writing, such as a power failure, another controller starts the Failover handler, mirroring the data in the write buffer of the fault controller to the local write cache to prevent Data loss, taking over the business that the faulty controller is running, does not affect the normal use of the user.。
The storage administrator can turn off the image caching mechanism as needed to further improve system performance. Of course, this will result in a decrease in system security and business continuity.。