Introduction to Networking for NonStop NS-Series Servers (H06.03+, J06.03+)

Networking on Integrity NonStop Servers
Introduction to Networking for HP Integrity NonStop NS-Series Servers—529874-003
1-7
Integrity NonStop Server Networking Compared to
Other Platforms
filtering and benefit in two ways from the parallel TCP/IP architecture. First, you could
run multiple httpd server processes (the listening processes) each bound to the same
port and IP address; this action scales your httpd server process up to 16 processors.
Second, no interprocess hop occurs from the processor containing the listening
process and a processor with TCP/IP access to the communication adapter. (Note that
the second benefit is not dependent on a round-robin filtering environment and is
available whenever you use NonStop TCP/IPv6.)
Additionally, the NonStop TCP/IPv6 architecture allows direct access from each
processor in a node to the adapter. (Direct access to the adapter is distinct from the
architecture of conventional TCP/IP (NonStop TCP/IP) which can involve an
interprocess hop from a processor containing the TCP/IP stack to the processor
containing the application.)
For more information about the parallel TCP/IP architecture available on NonStop
servers, about the models for listening applications and how they benefit from this
architecture, and for procedures for configuring the networking environment for round-
robin filtering, see the TCP/IPv6 Configuration and Management Manual.
For information about the iTP Secure WebServer, see the iTP Secure WebServer
System Administrator’s Guide.
The remainder of this subsection details the unique characteristics of networking in the
NonStop server environment.
You can use various methods for scaling your DNS implementation, some of which are
unique to the NonStop server and some of which are enhanced by the NonStop server
architecture. You can also use various methods for scaling your network interface
capacity as well as scaling your application through DNS capabilities.
For the purposes of this discussion we make the distinction between:
•
Network Scalability
•
Application Scalability
Network Scalability
Network scalability refers to the use of multiple physical interfaces to accommodate
bandwidth requirements for which a single network interface is insufficient. You can
achieve network scalability by having multiple network interfaces on multiple hosts or
by having a multi-homed host (a single system that has multiple network interfaces).
You can make this form of scalability either explicit or transparent. If explicit, then both
ends of the connection simply accommodate multiple interfaces by specifying a list of
IP addresses.
To make network scalability transparent with a multi-homed host or with multiple hosts,
you can use the Domain Name System (DNS) and have the multiple interfaces share a
single, externally visible name. One way to have multiple interfaces share a name, is to
have a DNS name server process on each of the subnets and configure each of those