In order to support small customers and lab and trial
deployments, the PCA application can scale down to run on a small hardware
footprint. This section describes the smallest supported PCA DSR deployments.
A lab or trial system may not be required to support
in-service maintenance or have any hardware redundancy whatsoever. In the
smallest supported lab/trial PCA DSR, IPFE is not included because it does not
make sense to distribute ingress connections when there is only one DA-MP
server.
The NOAM and SOAM servers are also running in simplex
mode, meaning that no redundancy exists. In addition, the NOAM and SOAM are
virtualized on a single physical server to save hardware. The binding and
session SBR servers are also running in simplex mode, but must be configured to
host either a Policy Binding and Policy Session database. A single DA-MP hosts
all Diameter signaling. Signaling is not affected if one or both of the
(virtual) OAM servers happens to fail.
The configuration of the smallest viable commercially
deployable PCA DSR, illustrated in , has enough hardware redundancy to support
in-service maintenance:
- Two DA-MPs are required to survive server failures and
maintenance. These DA-MPs should be engineered at 40% load since in a failure
or maintenance situation, one Server will have to handle the load for both.
- Both binding and session SBR Servers pairs use the
Active/Standby redundancy model in order to support failures and maintenance.
- The NOAM/SOAM Server pair uses the Active/Standby
redundancy model in order to support failures and maintenance.
- Both NOAM and SOAM are virtualized onto a single pair
of physical servers. The NOAM instance is Active on one server and Standby on
the other. The SOAM instance is Active on one server and Standby on the other.
The smallest supported Mated Pair of PCA DSRs,
illustrated in
Figure 2-9,
has certain characteristics:
- The NOAM servers are deployed at Site 1 using
Active/Standby redundancy.
- The Site 1 SOAM servers are deployed at Site 1,
virtualized on the same servers with the NOAM servers. They, however, use the
Active/Standby/Spare redundancy model, with the Spare server deployed at Site 2
and virtualized on the same server with one of the Site 2 SOAM servers.
- The Site 2 SOAM servers are deployed at Site 2 using
the Active/Standby/Spare redundancy model. The Spare Site 2 SOAM server is
virtualized at Site 1 on one of the servers already hosting an NOAM and a Site
1 SOAM server.
- A Binding SBR triplet is deployed with two servers at
Site 1 and one server at Site 2.
- A Session SBR triplet is deployed with 1 server at Site
1 and two at Site 2
- Two DA-MPs are deployed at each site to support server
redundancy at each site.
The smallest supported mated triplets of PCA DSRs,
illustrated in , has certain characteristics:
- The NOAM servers are deployed at Site 1 using
Active/Standby redundancy.
- The Site 1 SOAM server are deployed at Site 1,
virtualized on the same servers with the NOAM servers. However, the use the
Active/Standby/Spare/Spare redundancy model, with the spare server deployed at
Site 2 and Site 3 and virtualized on the same server with one of the Site 2
SOAM servers.
- The Site 2 (and Site 3) SOAM servers are deployed at
Site 2 (and Site 3) using the Active/Standby/Spare/Spare redundancy model. The
spare Site 2 SOAM server is virtualized at Site 1 on one of the servers already
hosting an NOAM and a Site 1 SOAM server.
- A binding SBR triplet is deployed with two servers at
Site 1 and one server each at Site 2 and Site 2 respectively.
- A session SBR triplet is deployed with two servers at
Site 2 and one server each at Site 1 and Site 3 respectively.
- Two DA-MP servers are deployed at each site to support
server redundancy at each site.
