ProLiant High Availability: The IT Imperative

monitor for connectivity to the core
network. The primary path automatically
fails over to the secondary path as soon as it
senses a loss of connectivity.
Fast path failover determines the fastest
path to the core switch to help maximize
network performance and availability by
identifying network path degradation.
For example, fast path failover would detect
if the Gigabit Ethernet connection from an
access switch to a core switch fails and the
traffic is re- routed over a 100-Mbps
Ethernet connection. It would then fail the
traffic over to an alternative Gigabit
Ethernet connection.
Dual channel teaming allows system
administrators to create a team of NICs
that support receive and transmit load
balancing, and provides switch redundancy.
This combination of capabilities is not
available with any other basic team types
such as switch-assisted load balancing (SLB)
or transmit load balancing (TLB). With dual
channel teaming, two NIC teams appear as a
single connection to the server. If one of
the switches fails, there is no loss of
connectivity and the failed switch can be
replaced without affecting server traffic.
Virus throttle can slow down the
spread of virus, leaving time for IT
intervention before the entire network
becomes infected. Traditional approaches
to anti-viral protection are based on the
actual code or signature of the virus. Virus
throttle identifies viruses on the basis of
their behavior. Because it works without
knowing anything about the virus, it can
handle unknown virus without waiting for
signature updates.
virtual
presence
HP’s Integrated Lights-out (iLO) technology
reduces cost and increases server
availability by giving an IT organization a
virtual presence within the data center as
well as on any remote system. That means
no matter where the server is located, the
IT organization has control over the key
system resources such as the console,
keyboard, mouse and power. Using iLO,
an IT organization even has the ability to
make storage media appear local to the
server. In addition, iLO continues to
operate even if the server’s operating
system is not functioning.
IT organizations can use iLO to
install, configure, monitor, update and
troubleshoot remote ProLiant servers from
a standard web browser, command line
or script without requiring any additional
software on the client system. iLO is
integrated with other management tools,
making it easier to combine virtual
presence capabilities with other server
lifecycle management tasks from
deployment to ongoing administration.
memory protection
technologies
ProLiant servers use a variety of techniques
to protect against errors, and hence
increase availability. For example, HP was
one of the first companies to introduce
advanced memory protection technology
such as ECC (Error-Correcting Code)
memory, online spare memory, mirrored
memory and RAID memory in industry-
standard servers.
To improve memory protection even
further, HP introduced Advanced ECC
technology. Advanced ECC technology is
capable of correcting a multi-bit error that
occurs within one dynamic random access
memory (DRAM) chip.
The ProLiant server online spare
memory determines if an active DIMM
(dual inline memory module) exceeds a
predefined error threshold. The error will
be corrected and the data from the entire
bank that contains the failed DIMM will be
copied to online spare memory. The failed
bank is deactivated, but the server will
remain available until the failed DIMM is
replaced during a scheduled shutdown.
Whereas online spare memory mode
protects against single-bit errors and entire
DRAM failure, mirrored memory mode
enables full protection against single-bit
and multi-bit errors. In mirrored memory
mode, the same data is written to both
system memory and mirrored memory
banks, but data is read only from the
system memory banks. If a DIMM in the
system memory banks experiences a
multi-bit error or reaches the pre-defined
error threshold for single-bit errors, the
roles of the system and mirrored memory
banks are reversed.
HP is one of the first companies to
support hot plug RAID memory, which
allows the memory subsystem to operate
continuously even in the event of a
complete memory device failure. In this
context, RAID stands for Redundant Array
of Industry-standard DIMMs.
Hot plug RAID memory generates
parity for an entire cache line of data
during write operations and records the
parity information on a dedicated parity
cartridge. However, hot plug RAID
memory does not have the mechanical
delays of seek time, rotational latency and
bottlenecks associated with disk drive
arrays.
figure 2. fast path failover
S P E C I A L A D V E R T I S I N G S E C T I O N
3a
FAILURE: team’s primary NIC has 100 Mbps path to core switch
CORE
SWITCH
HP ProLiant server
NIC 1
NIC 2
primary
secondary
Gigabit Ethernet 100 Mbps
▼
▼
RECOVERY: team’s new primary NIC has gigabit path to core switch
CORE
SWITCH
HP ProLiant server
NIC 1
NIC 2
new
primary
new
secondary
Gigabit Ethernet 100 Mbps
▼
▼
SWITCH A
SWITCH B
SWITCH A
SWITCH B
slower cross-link