Huawei quidway s7700 Configuration Manual page 236

Quidway S7700 Smart Routing Switch
Configuration Guide - SPU
Server Load Balancing
With the fast development of the Internet and services, the network-based data access traffic
increases rapidly. In particular, the traffic of access to data centers, large enterprises, and portal
websites reaches 10 Gbit/s. In addition, servers provide rich information for access users through
applications such as Hypertext Transfer Protocol (HTTP), File Transfer Protocol (FTP), and
Simple Mail Transfer Protocol (SMTP). The servers are flooded by the data gradually. Besides,
most websites especially e-commerce websites provide the around-clock non-stop service. In
this case, any service interruption or key data loss in communication will result in commercial
loss. These require high performance and high reliability for application services.
With the development of network technologies, the server processing speed and memory access
speed cannot meet requirements of network bandwidth and application service increase. The
network bandwidth increase brings in the increase of users, whereas server resources are
seriously consumed. The servers become the network bottleneck. Simply upgrading the
hardware of servers is expensive and is of poor extensibility. The problems such as single-point
faults on networks cannot be solved.
By using the dynamic load balancing algorithm, the server load balancing technology properly
allocates network services to servers in a server group. This reduces the burden of a single server
and improves the reliability of the server. You only need to add a server to a server group, without
changing the existing network structure or stopping existing services.
In server load balancing, the forwarding mode is classified into DNAT and DMAC. The process
of the two modes is the same. The SPU provides a virtual IP address. After users request services
through the virtual IP address, the SPU allocates the requests to real servers according to the
load balancing algorithm. The differences between the processing modes are described as
follows: In DNAT mode, when allocating service requests, the SPU translates the destination
IP address of service request packets, namely, the IP address of a real server, and forwards them
to real servers through routes. In MAC mode, when allocating service requests, the SPU replaces
the destination MAC address of the service request packets with the MAC address of a real
server without changing the destination IP address. Then the SPU forwards the packets to the
real server.
l
Issue 01 (2011-07-15)
The forwarding mode can be DNAT or DMAC in server load balancing. In egress link load
balancing, the SPU supports the redirection mode.
Server load balancing in DNAT mode
In DNAT mode, the networking is flexible. The backup servers can be located in different
physical positions and on different LANs.
Huawei Proprietary and Confidential
Copyright © Huawei Technologies Co., Ltd.
6 Load Balancing Configuration
Figure 6-2 shows the typical networking.
225