Talking about the Network Deployment Model of the Double Center in the Same City
When an enterprise builds a data center, two or more data centers will be built for disaster recovery considerations. For example, we often refer to the "three centers of the two places", that is, the same city double center + off-site center.
The same city dual center refers to the establishment of two data centers that can independently undertake business in the same city or neighboring cities. The dual centers have basically the same business processing capabilities and synchronize data in real time through high-speed links. In daily situations, the business and management systems can be shared simultaneously. The operation, and can be switched to run; emergency backup in case of disaster, to ensure the continuity of the business. The remote disaster recovery center refers to the establishment of a backup disaster recovery center in a remote city for data backup of the dual center. When a disaster occurs in a double center due to natural disasters, etc., the remote disaster recovery center can use the backup data for business. restore.
Today, I mainly talk about the network deployment model, switching methods, advantages and disadvantages of the same city dual center.
There are two deployment methods for the dual-center in the same city, which are mainly active and active. You can visit SPOTO to get more information.
1. Active and standby center model
From the literal point of view, the main and standby center is to build two data centers, one master and one backup, both data centers have the ability to independently provide external services. Under normal circumstances, the main center provides external services. When a major fault occurs, the main center's business is switched to the standby center, and the backup center becomes a new production center to provide external services.
The above description of the master-slave relationship is very simple, but in the actual deployment process, network, application, database, and other departments need to coordinate and cooperate to switch. We mainly talk about the network today.
Schematic diagram of the active and standby data centers:
Ac.pngac.png
The figure above is a typical network model of the main city center in the same city. The center A is the production center, the center B is the standby center, and the business divisions of the two centers A and B use the same IP address segment.
The technical key points of this model:
1. The network core (gateway) of the A and B centers are opened on the second floor;
2. The gateway is enabled on the main center side, and the standby center gateway is in the off state;
3. The primary center A advertises routes to the outside, and the B center does not advertise routes.
Switch between data centers:
The A center stops issuing routes to the outside, and the B center switches the gateway of the A center to the B center at the same time, thus completing the switching of the data center network.
Advantages of the master and backup models:
1. Simple technology and convenient switching;
2. The existing dual-center transition to this model is relatively simple and does not involve the application IP address transformation;
Main and standby model disadvantages:
1. There are many inter-center traversal traffic, and it is impossible to “the central server enters and exits from the center” (the gateways on both sides cannot be opened at the same time, otherwise the firewall asynchronous routing problem will occur).
2. The physical distance between the centers is too long, resulting in increased network delay, and delay-sensitive applications may be affected;
3. When the center L2 is opened, the Layer 2 domain will be expanded. One central Layer 2 fault may affect another center. Data center STP isolation or special technology, such as OTV, is required.
4. Consider the issue of how firewall policies are synchronized;
5. The link is idle and the resources are not maximized, such as the external WAN link of the B center.
2. Double live center model
Double-active data centers, that is, two data centers have the ability to provide services simultaneously. As shown below:
B.png
The two data centers A and B use separate IP address segments and provide services to the outside. The client relies on the DNS analysis results to determine which data center to enter.
The technical key points of this model:
1. The two centers use separate IP address segments, and all routes are advertised.
2. Use DNS to direct client traffic to the corresponding data center.
Data center switching:
1. Under normal circumstances, two data centers provide services at the same time;
2. When the data center fails on one side, change the DNS resolution to stop the service in the data center.
Double live data center advantages:
1. The two data centers are completely independent and do not interfere with each other;
2. Less traffic between data centers;
3. There is no need to consider the firewall asynchronous routing problem;
4. The link is more efficiently utilized and the economy is maximized.
Disadvantages of the double live data center:
On the technical side, there are no shortcomings. Feasibility, from the existing dual-center to the double-active data center, in addition to the IP address transformation, it also involves the application of DNS transformation, involving inter-departmental collaborative work, driving difficulties, this model is more suitable for new data centers.
Choosing a data center network model requires investigating the deployment requirements of the business/application/system/server, and considering the top-down unified considerations, the best for you is the best.
Remarks: Whether it is active or active, it is not difficult to achieve various "lives" at the network level. In the production environment, the biggest bottleneck currently faced is actually the "double live" problem of the dual-center database.
The same city dual center refers to the establishment of two data centers that can independently undertake business in the same city or neighboring cities. The dual centers have basically the same business processing capabilities and synchronize data in real time through high-speed links. In daily situations, the business and management systems can be shared simultaneously. The operation, and can be switched to run; emergency backup in case of disaster, to ensure the continuity of the business. The remote disaster recovery center refers to the establishment of a backup disaster recovery center in a remote city for data backup of the dual center. When a disaster occurs in a double center due to natural disasters, etc., the remote disaster recovery center can use the backup data for business. restore.
Today, I mainly talk about the network deployment model, switching methods, advantages and disadvantages of the same city dual center.
There are two deployment methods for the dual-center in the same city, which are mainly active and active. You can visit SPOTO to get more information.
From the literal point of view, the main and standby center is to build two data centers, one master and one backup, both data centers have the ability to independently provide external services. Under normal circumstances, the main center provides external services. When a major fault occurs, the main center's business is switched to the standby center, and the backup center becomes a new production center to provide external services.
The above description of the master-slave relationship is very simple, but in the actual deployment process, network, application, database, and other departments need to coordinate and cooperate to switch. We mainly talk about the network today.
Schematic diagram of the active and standby data centers:
Ac.pngac.png
The figure above is a typical network model of the main city center in the same city. The center A is the production center, the center B is the standby center, and the business divisions of the two centers A and B use the same IP address segment.
The technical key points of this model:
1. The network core (gateway) of the A and B centers are opened on the second floor;
2. The gateway is enabled on the main center side, and the standby center gateway is in the off state;
3. The primary center A advertises routes to the outside, and the B center does not advertise routes.
Switch between data centers:
The A center stops issuing routes to the outside, and the B center switches the gateway of the A center to the B center at the same time, thus completing the switching of the data center network.
Advantages of the master and backup models:
1. Simple technology and convenient switching;
2. The existing dual-center transition to this model is relatively simple and does not involve the application IP address transformation;
Main and standby model disadvantages:
1. There are many inter-center traversal traffic, and it is impossible to “the central server enters and exits from the center” (the gateways on both sides cannot be opened at the same time, otherwise the firewall asynchronous routing problem will occur).
2. The physical distance between the centers is too long, resulting in increased network delay, and delay-sensitive applications may be affected;
3. When the center L2 is opened, the Layer 2 domain will be expanded. One central Layer 2 fault may affect another center. Data center STP isolation or special technology, such as OTV, is required.
4. Consider the issue of how firewall policies are synchronized;
5. The link is idle and the resources are not maximized, such as the external WAN link of the B center.
2. Double live center model
Double-active data centers, that is, two data centers have the ability to provide services simultaneously. As shown below:
B.png
The two data centers A and B use separate IP address segments and provide services to the outside. The client relies on the DNS analysis results to determine which data center to enter.
The technical key points of this model:
1. The two centers use separate IP address segments, and all routes are advertised.
2. Use DNS to direct client traffic to the corresponding data center.
Data center switching:
1. Under normal circumstances, two data centers provide services at the same time;
2. When the data center fails on one side, change the DNS resolution to stop the service in the data center.
Double live data center advantages:
1. The two data centers are completely independent and do not interfere with each other;
2. Less traffic between data centers;
3. There is no need to consider the firewall asynchronous routing problem;
4. The link is more efficiently utilized and the economy is maximized.
Disadvantages of the double live data center:
On the technical side, there are no shortcomings. Feasibility, from the existing dual-center to the double-active data center, in addition to the IP address transformation, it also involves the application of DNS transformation, involving inter-departmental collaborative work, driving difficulties, this model is more suitable for new data centers.
Choosing a data center network model requires investigating the deployment requirements of the business/application/system/server, and considering the top-down unified considerations, the best for you is the best.
Remarks: Whether it is active or active, it is not difficult to achieve various "lives" at the network level. In the production environment, the biggest bottleneck currently faced is actually the "double live" problem of the dual-center database.
评论
发表评论