Effects of virtualization and cloud computing on data center networks

13
IRF lets you create large network fabrics with multiple switches at a single layer (access, distribution,
or core) that operate and appear logically as a single switch. You can combine as many as eight HP
networking stackable switches or up to four HP networking chassis switches to create a single, logical
switch comprised of hundreds or even thousands of 1-GbE or 10-GbE switch ports. Because you now
have a single switch, the routing protocols calculate routes based on a single logical domain rather
than the multiple switches it represents.
IRF mimics the management behavior of a chassis switch by moving what was the management
control plane in the backplane of a chassis switch to a distributed control plane across the network
fabric. You only need one configuration file and one software image to manage IRF devices. Devices
inserted into an IRF domain automatically update their configuration file and software, preventing you
from modifying one device in the domain in isolation from the others. Connecting to the fabric
through any port, console port or management port will link you to the single, redundant domain
controller hosted on an elected IRF device in the domain.
Edge or aggregation switches (including Virtual Connect) that are dual-homed to IRF-enabled switches
effectively “see” the associated upstream switches as a single entity, eliminating the need for slow
convergence technologies such as STP. IRF-enabled switches also let you use simple link-aggregation
protocols (such as 802.3AD LACP) for effective, failure tolerant active-active multi-chassis dual
homing. This creates a distributed switch that is highly available, has no single point of failure, and
requires no complex load balancing or failover protocols.
Unlike STP, which limits the active links to prevent loops, IRF delivers high performance by fully using
all links between switches and servers. IRF also provides low latency communications, ensuring rapid
network recovery (less than 50 milliseconds) from a link or device failure. This is much faster than the
several seconds that an STP- or even an RSTP-network uses.
Because IRF is a switch virtualization architecture, it works with other higher layer protocols rather
than competing with them. This includes protocols like Shortest Path Bridging (SPB), IETF Transparent
Interconnection of Lots of Links (TRILL), and Multiprotocol label switching (MPLS). This gives you the
most flexibility while maintaining resiliency and performance.
IRF runs on the high-end HP networking switches. The HP 12500, 9500, 7500, 58XX, and 55XX
switches all come with HP IRF technology built in.
HP IRF with HP Virtual Connect technology
By combining IRF with Virtual Connect technology, you can take an existing three-tier network and
flatten it to two tiers—the aggregation layer with IRF and the network core layer (Figure 8). You also
get the advantages of reduced hardware at the server edge by using VC FlexFabric adapters and
modules.