Understanding and Designing Serviceguard Disaster Recovery Architectures

Table 7 Comparison of disaster recovery Cluster Solutions (continued)
ContinentalclustersMetroclusterExtended Distance Cluster
for RAC
Extended Distance
Cluster
Attributes
many scenarios that have
multiple failures).
re-sync is required in
many scenarios that
have multiple failures.)
solutions (including
Continuous Access XP,
Continuous Access
subsystem, so the host
does not experience a
performance hit.
EVA, 3PAR RemoteIncremental re-syncs are
Copy and EMC SRDFdone, based on bitmap,
for array-based, orminimizing the need for
Data Replicationfull re-syncs. Site Aware
Services toolkitDisaster Tolerant
provided byArchitecture (SADTA)
Serviceguard toprovides a method to
integrate Oracle Dataconfigure a disaster
Guard and IBM DB2recovery solution for
HADR replication
technologies.
complex multi-instance
workloads such as Oracle
RAC databases.
Semi-automatic (Allows
users to “push the
Automatic (no manual
intervention required).
Instance is already
running at the second site.
Automatic (no manual
intervention required).
Application
Failover
button” to initiate
recovery).
Data Replication Storage
Failover Preview
Disaster Recovery (DR)
Rehearsal provides a
method to identify and
fix configuration
inconsistency at the
recovery cluster. See
(see “Disaster Recovery
Rehearsal” (page 34).
Active/StandbyActive/StandbyActive/ActiveActive/StandbyAccess Mode
per
Application
You must reconnect
after the application is
Client detects the lost
connection. You must
Client may already have
a standby connection to
remote site.
Client detects the lost
connection. You must
reconnect after the
Client
Transparency
recovered at second
site.
reconnect once the
application is recovered
at second site.
application is
recovered at second
site.
Based on the limit of the
number of nodes per
3 to 16 nodes* 2 nodes with SLVM or
CVM with a maximum
2 to 16 nodes (up to
four when using dual
Maximum
Cluster Size
Allowed Serviceguard/SGeRAC
cluster.
distance of 100km. * 2,
4, 6, or 8 nodes to eight
lock disks) using LVM
or VxVM. 2, 4, 6, or
nodes with CVM or CFS8 nodes (up to four
with a maximum distance
of 10km.
when using dual lock
disks) using CFS.
Identical storage is
required if
Identical Storage is
required.
Identical storage is not
required (replication is
Identical storage is not
required (replication is
Storage
storage-based mirroring
is used.
host-based with either
*MirrorDisk/UX or CVM
Mirroring).
host-based with either
MirrorDisk/UX or
VxVM mirroring).
Identical storage is not
required for other data
replication
implementations.
WANDark FiberDark FiberDark FiberData
Replication
Link
LAN
Dark Fiber
(pre-integrated solution)
Continuous Access over IP
Continuous Access over
ATM
71