Table of Contents
Advertisement
– Utilize the CPU in each grid controller, as much as possible, for high CPU
utilization. To achieve this, it is recommended to use ports 1 and 4; and then
ports 2 and 3.
v For high availability in each host:
– Divide the paths between the two switches/fabrics: half the paths connecting
to switch 1 and the other half connecting to switch 2.
– If possible, spread the paths across different grid controllers.
– Use different adapters within each of the grid controllers.
Important: It is important to cross the Fibre Channel or iSCSI ports for resiliency
and high performance, as depicted in Figure 29.
Figure 29. Crossing Fibre Channel grid controller ports
For Fibre Channel port numbering on the rear of the grid controllers see Figure 30
on page 68.
For Ethernet port numbering on the rear of the grid controllers see Figure 31 on
page 69.
Be sure that half of the ports in each grid controller connect to the first switch and
half of the ports of each grid controller connect to the second switch.
Important: Non-redundant configurations, such as single fabric/zone and
single-port host configurations may be implemented, however they are not
recommended.
Use non-redundant configurations only when the risks of a single point of failure
are acceptable, which is typically the case for test and development environments.
Fibre Channel (FC) network configurations
Host systems can connect to the storage system over a Fibre Channel (FC)
network.
In a production environment, always connect Fibre Channel hosts to a minimum of
two separate SAN switches in independent fabrics to provide redundancy. In Fibre
Channel network configurations, there are two dual-port connections per grid
controller.
1
3
2
4
Chapter 5. Network and host connectivity requirements
67
Hide quick links:
Advertisement
Table of Contents
