Use LinuxonPower blade server for complex networks

Article title: use LinuxonPower blade server to implement a complex network. Linux is a technology channel of the IT lab in China. Includes basic categories such as desktop applications, Linux system management, kernel research, embedded systems, and open source.

Blade servers are the best choice for many applications and services, especially for telecom service providers. However, the unique requirements of these providers' networks often require complex configurations and require detailed research and planning in advance to meet all strict functional requirements. Through this article, we can learn how to deploy POWER6? How to plan and set up necessary network configurations for JS22 blade servers.
For the following reasons, the operation model based on blade servers is of great value in the wired and wireless telecommunication fields:

1. the memory usage is small, and the data center space can be used more effectively.
2. it can meet the NEBS needs of distributed deployment. (NEBS [Network Equipment Building System] is a set of conditions that a Network connection device must meet in order to achieve compatibility .)
3. it provides cost-effective horizontal scalability for telecom service providers to reduce deployment costs.
4. provides centralized management support to provide better OAM & P support for service provider network deployment. (OAM & P stands for "operations, administration, maintenance and provisioning ". This term represents the principles and software required for management ).
5. built-in support for continuous availability-based operation models (including upgrade and maintenance activities) to avoid service downtime that users can perceive.
In addition, the following factors are also important in the telecom service provider environment, especially in environments with complex configurations:

Multiple VLANs. These are used for CDN (Customer Data Network) and Management (OAM & P) communication streams. Consider them separately to ensure that QoS (service quality) is effectively maintained across multiple LPAR (logical partitions ).
Differential zone and virtualization. These strategies help maximize capacity utilization and reduce TCO (total cost of ownership ).
Complexity of existing networks. The load of the existing network may change greatly, so load balancing must be performed between multiple client LPAR.
This article describes how to use a combination of active and passive Cisco switches to implement multi-VLAN configuration for a blade server rack. In our example, how does the configured network connect to a Linux instance? On Power BladeCenter? Multiple VLANs on JS22. This architecture consists of six Cisco catalyst switch modules, each of which has 14 internal ports and 4 External 1 GB ports.

To properly utilize all six switches in the rack, the blade server requires six Ethernet interfaces. The Ethernet interface ent0 on the blade server maps to the first switch on the blade server rack. Each Ethernet interface maps to the next available switch in turn. This ing method has a limitation because it does not allow administrators to map physical adapters on the blade server to the switches they select.

Each Cisco switch on the rack must have a physical interface when creating a network architecture for the blade server. If the number of adapters on some blade servers is different from the number of adapters on the rack, the blade server will not be able to use the switches that are not associated with the blade server physical adapter.

Be sure to know the Ethernet interface on the blade server and the matching method of the switch in the rack. The first switch in the rack is usually placed in the upper left corner of the blade server rack, right under the power plug. It maps to ent0 on the blade server because it is the first interface on the blade server.

Figure 1 shows the switch number in our configuration.

It is extremely important to determine the vSwitch pairing mode to achieve high availability. In a typical configuration, one power distribution unit (PDU) powers the upper half of the blade server rack, and the other PDU powers the lower half of the rack. To create a redundant solution, you must place the primary switch and secondary switch on the upper half and lower half of the rack respectively.

In our example, the created solution is paired with the following vSwitch: (1, 9) (2, 3) (4, 7 ). Because the adapter pair (ent0, ent1), (ent2, ent3), and (ent4, ent5) are on the same physical I/O card, therefore, make sure that the network communication flow of the target VLAN does not pass through the same I/O card. Our configuration requires communication flow distribution across multiple PDUs and multiple interfaces.

Although the pairing of the adapter and switch looks simple, you need to perform multiple steps to configure IVM (Integrated Virtualization Manager), switch, and LPAR to use this architecture. Figure 2 shows the configuration of a blade server and associated switches, aggregation (trunk), and VLAN tags. This configuration enables multiple VLANs to communicate with multiple switches through the same physical adapter.

Figure 2. a blade server and associated switches, aggregation, and VLAN tags

 

In this example, each LPAR has two Ethernet adapters connected to a Virtual Ethernet Adapter (VEA) on the IVM ). Note that VEA on IVM is associated with multiple VLANs. The communication of each VLAN passes through the distribution of their adapters to the LPAR. VEA integrates VLAN communication streams, sends them through a link aggregation device through the Shared Ethernet Adapter (SEA), and sends them out through one of the rack switches. A vSwitch uses VLAN tags to route VLAN communication streams to an appropriate network.

[1] [2] [3] [4] [5] Next page