At present, with the advancement of enterprise informatization construction, more and more enterprises choose to build their own mail systems as information transfer hubs and storage archives. The employee's reliance and trust in the mail system also brings potential hidden dangers, that is, if the server of the system is down or the service is stopped due to uncontrollable factors, there will be incalculable consequences.
First, let's take a look at the situations that plague the communication of enterprises. Specifically, there are the following types:
1. Frequent failures such as power outages;
2. The Internet infrastructure construction where the server is located is seriously backward;
3. The problem of network interoperability between the two places (for example, there is a disagreement between China Southern Telecom and China Unicom);
4. The bandwidth of international ingress and egress is unstable and the delay is large;
5. Firewall interference between countries;
6. When all users access the headquarters server centrally, local imbalance will be caused, which will bring great pressure to the network and server.
After such a situation occurs, the processing speed of the system will be affected at light, and data will be lost at worst. U-Mail, a well-known domestic brand, suggests to change the situation of low efficiency and high risk of a single server architecture, prepare several more servers for distributed deployment, and distribute the load. Its principle is:
1. A set of mail servers are deployed in each computer room node to respond to the mailbox service requests in their respective regions, to solve the slow system response caused by the difference in the network environment of the computer room nodes, to respond to the mailbox service needs of their respective areas, and to disperse the headquarters at the same time Server access pressure;
2. Each server can send and receive emails with the external network;
3. A server can be randomly selected to receive external network mail, or set according to the server priority;
4. When a server receives mail, if the recipient does not exist, it will automatically find another server, and if it does not exist, it will return the mail;
5. When an employee is transferred from the headquarters to a branch, the account and email data information can be automatically migrated to the branch server, and the servers in each region can be redundant.
Therefore, the U-Mail mail system provides a remote distributed + dual-machine hot backup solution. Its network topology is as follows:
As shown in the figure above, U-Mail mail servers are set up on three nodes: workstation, computer room A, and computer room B to form a distributed solution. Two servers are deployed in each area to form a dual-system hot backup solution.
The distributed + dual-computer hot standby solution can solve the load balancing and data active-active hot backup requirements of the workstation, the A computer room, and the B computer room. The load balancing is solved by the u-mail distributed deployment solution, and the data active-active hot standby adopts dual computers Hot standby HA solution.
In order to prevent the failure of a single server and stop the service, some companies will arrange manpower to monitor the server, detect abnormalities and then deal with it, it is too late, it is better to prevent problems before they occur, because:
1. After the failure, it needs to be handled manually, which will increase the labor cost;
2. It takes time to deal with the fault, including the staff arriving at the scene, screening the problem, confirming and eliminating the problem;
3. Some problems are quite difficult, such as server hardware failure, operating system failure, etc., which cannot be handled well in a short time;
4. It is impossible for managers to monitor 24 hours a day, but the failure does not know when it occurs.
Therefore, enterprises need U-Mail's dual-system hot-standby solution, which connects two servers through a data cable and automatically detects it. Once a server fails, it can quickly switch to another server and continue to provide services. the whole process:
1. Fully automatic, without manual intervention;
2. The switching is very fast, almost imperceptible;
3. The email account and data are synchronized, or stored in another storage device;
4. No data loss, to ensure stable output.
The network topology of the dual-system hot backup solution is as follows:
The two servers share a disk array, provide external services with a virtual IP address, and send external service requests to one of the servers to undertake. The server detects the working status of the other server in real time through the heartbeat line. Once it detects an abnormality or a failure, the other server quickly judges and automatically switches.
Application Environment:
The dual-machine hot backup solution is suitable for enterprises or units with the following needs:
1. Requires the uninterrupted service of the mail server;
2. The mail server needs to provide high-speed, reliable and stable services;
3. Data uniqueness and high security are required;
4. There is a need to keep the business running continuously.
The above is to solve the problem that the local server is prone to failure, and to achieve the synchronization of the communication between the branch and the headquarters, a distributed deployment can be carried out. The schematic diagram is as follows:
1. Deployment environment
The servers in the three regions are installed with the U-Mail mail system. The domain name must be the same, with comingchina.com as the column (please replace it with your own domain name). Each server creates an MX record and points to the corresponding public IP address. Each server account cannot be the same. If a duplicate account is added, a conflict will be automatically prompted.
2. Receive external emails
Emails sent from the external network are selected and delivered to a certain server according to the priority of the MX records of the domain name.
MX1: mx1.comingchina.com
MX2: mx2.comingchina.com
MX3: mx3.comingchina.com
If the recipient is not on the current receiving server, it will automatically check the other two servers through the relevant TCP protocol, and if the other two servers have the recipient, the mail will be transmitted. If not, bounce back to sender.
3. Send and receive emails internally
Internal sending and receiving use the servers in their respective regions to send and receive mail. If the recipient is not on the current server, it will automatically check the other two servers through the relevant TCP protocol. If the other two servers have the recipient, the mail will be transmitted. If not, bounce back to sender.
4. outgoing mail
Outgoing mail is sent using the server in the respective region.
5. Automatically forward mail
For a single server, sometimes some mail delivery fails. At this time, you can use the resources of multiple distributed servers to set up the automatic TCP protocol connection to another server to continue the delivery when a server fails to deliver, increasing the successful delivery rate.
6. Automatically migrate account data
When an employee is transferred from the headquarters to a branch, the account, email data and other information can be automatically migrated to the branch server through the TCP protocol. There is no need to delete from one server, and then create another server. Humanized experience, providing work efficiency.
7. High availability
Two servers are deployed in each region to form a high-availability (dual-system hot backup) solution. When one server fails, it automatically switches to another server.