Bull Escala EPC400 Connecting Manual page 150

Rack 400:
Rack 1200:
Disk Drives:
Attachment:
Fibre Channel Hub
The FC–AL hub of Escala offer is the Gadzoox's FCL1063TW hub. The FCL1063TW is an
active hub for Fibre Channel Arbitrated Loop storage clusters. It has intelligent, per–port
data detection check capabilities for automatic port bypass control. The FCL1063TW is
configured to have 9 default copper ports each of which are supplied with power to support
the Fibre Channel media Interface Adaptors (MIA) for electrical to optical conversion. The
FCL1063TW operates without any user intervention. It continuously monitors the serial data
and automatically configures an FC–AL network based on its intelligent bypass control logic.
To operate, one adds valid FC–AL nodes with ANSI compliant cables – plug and play
operation.
Regarding Dual Loop Configuration, two hubs are recommended. The FCL1063TW is
designed as a precise half rack chassis so that full redundancy can be achieved with a
single, 1U height rack space. Two hubs used together for dual loop applications provide
dual power supplies and independent loop ports, and allow for field repairability without
system down time.
The FCL1063TW can be mounted into a rack. There is a double rack mount kit designed
for mounting two hubs, side by side into a rack.
There is no power switch on the Fibre Channel Hub. The hub should be the first operating
component, before powering up the nodes with FC–AL adapters and the DAS fibre
(DAS3500, DAS57xx or DAS5300).
The removal of an SP on a DAS57xx and DAS5300 breaks (no by-pass) the FC-AL loop.
Therefore, the two SPs of a DAS 57xx and DAS5300 cannot be linked between themselves.
The use of a Fibre Channel Hub is thus recommended for Dual-SP DAS 57xx and DAS5300
as pictured in SLOOP01/03.
Configuration for High-Availability and Disaster Recovery
HACMP in conjunction with the support of Fibre connection in DAS subsystems can be used
for data mirroring between systems placed in separate locations for extended high
availability and disaster recovery ( see XLOOP02 figures). The base configuration is made
of two PCI nodes (EPC400/440, EPC1200, EPC1200A or EOC2400 nodes) which are
located in two floors of the same building, or in two separate buildings within a campus. The
two nodes are equipped each with a system console or a graphic display. It is assumed that
there is a customer's public network (LAN) to which the nodes are connected. There is no
interconnect. A DAS fibre disk subsystem is attached to a node and its data are mirrored on
the remote DAS fibre disk subsystem attached to the second node. There are two hubs on
each side. A FC-AL hub is used to establish a path from a node to the local disk unit. Two
paths exist between a node and a disk unit.
Distances Up to 500 meters
Case 1 – IP takeover on node failure:
For implementing a HACMP cluster, the two nodes must be connected to a same subnet of
the customer's public network. The heartbeat serial line between the two nodes is extended
by means of micro-modems on each side.
10-40
EPC Connecting Guide
CKTG109-0000 – Rackmount option (DAE 5000)
CKTG110-0000 – Rackmount option (DAE 5000)
MSUG110-0F00 – 8.8GB Fibre DAE Disk (10 000rpm)
MSUG111-0F00 – 17.8GB Fibre DAE Disk (10 000rpm)
1 x DCCG141-0000 – PCI Fibre Channel Adapter
1 x DCCG147-0000 – PCI Fibre Channel Adapter
1 x DCCG148-0000 – PCI Fibre Channel Adapter
1 x FCCQ002-2000 – Cord 2CU/DB9 10M