Cisco Cisco Catalyst Blade Switch 3030 for Dell White Paper

Page of 29
 
Design Guide 
Aggregation Layer Switch Physical Port Count 
The introduction of blade systems into the data center requires greater port density at the 
aggregation layer. Blade systems, deployed with internal switches, provide their own access layer. 
The cabling and maximum number of servers per enclosure is predetermined. Scaling the 
aggregation-layer ports to accommodate the blade system uplinks is an area that requires 
attention.  
It is important to remember that aggregation switches provide data center services such as load 
balancing, security, and network analysis that may require dedicated ports for appliances or slots 
for integrated services. This directly affects the number of ports available for access-layer 
connectivity. 
Aggregation Layer Switch Slot Count 
The data center infrastructure must be flexible enough to allow growth in both server capacity and 
service performance. Connecting a blade system directly into the aggregation layer places more 
significance on the number of slots available to accommodate blade system uplinks and integrated 
services.  
Traditionally, the access layer provides the port density necessary to allow the physical growth of 
server farms. Modular access-layer switches offer connectivity to densely packed server farms over 
a few uplinks. The aggregation-layer switches support a limited number of uplinks from the access 
layer. With this model, the number of servers supported per uplink is high. 
Blade systems use more aggregation-layer resources per server than this traditional deployment 
model. Each uplink from a blade enclosure provides connectivity to a maximum of 16 servers. The 
aggregation layer must be flexible enough to manage the increased demand for ports and slots in 
this blade server system environment.  
To scale the server farm, use an aggregation-layer switch that provides an ample number of slots 
for line cards and/or service module expansion.  
In addition, consider using the following two options (which are not mutually exclusive): 
● 
Deploying service switches in the aggregation layer (as depicted in Figure 8) 
● 
Using a data center core to accommodate multiple aggregation-layer modules 
Service switches are deployed in the aggregation layer to host integrated data center services such 
as load balancing, intrusion detection, and network analysis. Relocating these services to a 
separate switch frees ports and slots in the aggregation-layer switches. This design allows the 
aggregation switches to commit more slots and, ultimately, more ports to the Layer 2 connectivity of 
the server farms.  
The data center core is a mechanism to replicate and horizontally scale the data center 
environment. In the recommended design the aggregation and access layers are regarded as a 
module that can be duplicated to extend the enterprise. Each data center module provides its own 
network services locally in the aggregation switches. This approach allows the network 
administrator to determine the limits of each data center module and replicate as necessary.  
Figure 6 depicts the data center design. The aggregation switches for each data center module are 
Layer 3 attached to the core. In addition, the aggregation switches house the service modules 
required to support the server farms.   The aggregation layer provides the connectivity layer for the 
 
© 2008 Cisco Systems, Inc. All rights reserved. This document is Cisco Public Information. 
Page 13 of 29