IBM HS21 Server User Manual


 
Workhorse 2-socket dual and quad-core
Intel Xeon blade server
Please see the Legal Information section for important notices and information.
8.
with the blade servers, using the same blade slots. Up to four chassis can be installed in an
industry-standard 42U rack, for a total of up to 56 30mm blade servers per rack.
Up to ten module slots for communication and I/O switches or bridges — The modules
interface with all of the blade servers in the chassis and alleviate the need for external switches or
expensive, cumbersome cabling. All connections are done internally via the midplane. Two module
slots are reserved for hot-swap/redundant Gigabit Ethernet switch modules. Two slots support
either high-speed bridge modules or legacy Gigabit Ethernet, Myrinet, Fibre Channel,
InfiniBand and other switch modules. Two slots are dedicated for bridge modules. Four additional
slots are dedicated for high-speed bridge modules. All modules, when in stalled in pairs, offer load
balancing and failover support.
Integrated switch and bridge modules mean that no additional rack “U” space is required.
Two module bays for Advanced Management Modules — The new management module
provides advanced systems management and KVM capabilities for not only the chassis itself, but
for all of the blades and other modules installed in the chassis. The Advanced Management Module
provides capabilities similar to the IBM Remote Supervisor Adapter II used in stand-alone xSeries
rack and tower servers. New features include concurrent KVM (cKVM) and media tray, an external
Serial over LAN connector, more memory, a more powerful onboard processor, industry-standard
management interfaces (SMASH/CLP/CIM/HPI), USB virtualization, network failover and backward
compatibility with the original Management Module, among others. The features of the module can
be accessed either locally or remotely across a network. One module comes standard. A second
module can be added for hot-swap/redundancy and failover. The module uses a USB
connection, rather than the PS2 connection of the original Management Module.
Two module bays for Blower Modules — Two hot-swap/redundant blower modules come
standard with the chassis. They are capable of providing efficient cooling for up to 14 blades. These
modules replace the need for each blade and switch to contain its own fans. The blowers are more
energy efficient than dozens or hundreds of smaller fans would be, and they offer many fewer points
of potential failure. BladeCenter H also includes up to twelve additional hot-swap/redundant fans to
cool the power supplies and high-speed switch modules.
Four module bays for Power Modules — BladeCenter H ships with two 2900W high-efficiency
hot-swap/redundant power modules (upgradeable to four), capable of handling the power needs
of the entire chassis, including future higher-wattage processors. Each power module includes a
customer-replaceable hot-swap/redundant fan pack (3 fans) for additional cooling capability.
A hot-swappable Media Tray containing a DVD-ROM drive, two USB 2.0 ports, and a light path
diagnostic panel — The media tray is shared by all the blades in the server. This reduces
unnecessary parts (and reduces the number of parts than can fail). In the event of a failure of the
Media Tray the tray can be swapped for another. While the tray is offline, the servers in the chassis
can remotely access the Media Tray in another chassis. The light path diagnostic panel contains
LEDs that identify which internal components are in need of service.
A serial breakout port with optional cable — This provides a direct serial connection to each
blade server installed in the chassis, as an alternative to Serial over LAN. (Note: This applies only to
newer blades that include this capability.)
It is extremely important to include all infrastructure costs when comparing a BladeCenter H
solution to a competitor’s offering, not just the cost of the chassis and the blades. The high density
and level of integration of the BladeCenter H chassis can greatly reduce the cost of the overall
solution. For example, because up to four chassis will fit in a rack, this means that up to 56 blade
servers can be installed. Also, because up to ten Ethernet, Myrinet, Fibre Channel, InfiniBand or
other bridges and switches can be installed per chassis, up to 40 switches and bridges can be
installed per rack without having to reserve any “U” space for switches, unlike the competition.
(And the integrated switches may be less expensive than external, self-powered switches.) Plus, the
number of power distribution units (PDUs) needed per rack may be lessened, because there
are fewer discrete devices to have to plug in. In addition, because all the blades are connected to all
the switches inside the chassis, there is no need for external Ethernet or other communication
cables to connect the blades, bridges and switches. (Only the few cables needed to connect the
switches to the external world are required.) This not only saves the cost of numerous cables per
rack, but also the clutter and bother of routing that many cables. An added bonus is potentially
much freer airflow behind the rack, due to fewer cables.
BladeCenter T is a carrier grade, rugged 8U (20-inch deep) chassis designed for challenging
central office and networking environments. It provides:
NEBS 3/ETSI-compliance — Designed for the Network Equipment Provider (NEP)/Service
Provider (SP) environment. Also ideal for government/military, aerospace, industrial
automation/robotics, medical imaging and finance.
Designed for Carrier-Grade Linux — Several distributions are supported, include SUSE and Red
Hat.
Reduced single points of failure — Many major components (either standard or optionally) are
hot-swappable and/or redundant. Servers and modules can be configured for automatic failover