Database Reference
In-Depth Information
Basically, stretch clusters are used for a very specific reason: Building Active-Active RAC clusters, spread
over a geographically spread-out location. However, due to network latency and cost issues, Active-Active Multi-
Master Golden Gate is a much more viable option than extended distance clusters. It is important to consider that
stretch clusters
are not
and
should not
be used as a replication/Disaster Recovery technology; Data Guard is the
recommended technology for DR/failover purposes.
Extended Distance (Stretch) Clusters: Setup/Configuration Best Practices
Following are good design best practices critical to the smooth operation of extended distance clusters:
•
Implement/configure redundancy at the public network layer
•
Dense wavelength division multiplexing (Dark Fibre) is the general technology of choice for
implementing a dedicated networking tier for stretch clusters, as it provides the high-speed
low-latency network channel needed for the private cluster interconnect
•
Implement and configure full redundancy at the private cluster interconnect networking layer
•
The private cluster interconnect must be on the same subnet across all sites/nodes
•
The public network must be on the same subnet across all sites/nodes
•
Implement/configure full redundancy at the storage fabric layer
•
Set up/configure ASM for normal redundancy for maintaining a local copy of the database at
all sites of the RAC cluster
•
Implement/configure preferred mirror disk reads (ASM_PREFERRED_READ_FAILURE_
GROUPS parameter) within ASM so that the disks nearest (local) to the particular nodes are
utilized
•
Implement the voting disk on a 3
rd
geographical site on NFS; this can be done over a WAN
•
Do not implement extended distance clusters over distances of more than 50-70 miles
Following are the pros/benefits/advantages of extended distance clusters:
•
Fast site-level failover protection solution among geographically spread-apart nodes
•
Active-Active configuration: full consumption/usage of all involved hardware
•
Scalability/High Availability RAC features not limited to a single geographical location
•
ASM data mirroring: multiple copies of the same data in geographically distant locations
•
Fast: good overall performance
By the same token, extended distance clusters are limited in their capabilities by the following
disadvantages/cons:
•
High costs associated with the need for high-speed networks
•
A dedicated redundant high-speed network is needed for the public/private network and
storage tiers. General rule of thumb: the more the geographical distance, the more the network
latency
•
Generally/usually, after 50-100 miles, the latency with most current technologies starts to gets
prohibitive for the smooth operation of an extended distance (stretch) cluster
•
No DR protection from large-scale geographical disasters due to the latency limitations
imposed by the need for a high-speed dedicated network infrastructure