Deployment options for Db2 pureScale

Db2 pureScale features four distinct deployment options, each addressing a specific high availability (HA) and/or disaster recovery (DR) requirement with an integrated Db2® solution with no additional dependencies on other software or storage solution.

Deployment models

The following Db2 pureScale deployment models are used for different user environments. Regardless if which model fits your needs, each of the Db2 pureScale deployment models share more than 95% of best practices in terms of system resource allocation (CPU and memory), network (switch, cluster interconnect), storage (disk types, Spectrum Scale configuration), Db2 instance, database, and client configurations.

HA in a single site

This deployment model is for customers who need continuous availability for only a single site. This specific cluster topology consists of multiple members, dual cluster caching facilities (CFs), multiple adapters and switches and a single storage accessible by all members and CFs. It provides constant availability and resiliency against planned and unplanned outages, unlimited capacity and workload balancing. Figure 1 shows this setup.
Figure 1. HA in a single site

HA in a single site with redundant storage

This deployment option is for customers looking for storage redundancy options in Db2 pureScale for a single site. This model is to deploy the Db2 pureScale with Spectrum Scale replication for a fully integrated solution that provides transient storage failover capability. Figure 2 shows this model.
Figure 2. HA in a single site with redundant storage

Geographically dispersed pureScale cluster (GDPC)

This deployment model is for users that require an active/active DR solution with zero down time in-site failure. Essentially, this model consists of a split Db2 pureScale cluster stretched over tens of kilometers. It can be deployed in different rooms or floors within the same building, between different buildings in the same city, or even different cities but within public transport range. This distance limitation within tens of kilometers is to ensure that the performance delay, due to the distance, is within an acceptable range for a reasonable workload. For more information on GDPC, refer to Geographically dispersed Db2 pureScale cluster (GDPC).

Db2 pureScale with HADR

If active/active DR is not a requirement, then the Db2 pureScale with HADR is another deployment model. This is very similar to HADR support in non-pureScale systems. It is worthwhile to emphasize that Db2 pureScale with HADR is not limited to large, geographic distance use cases, it can also be deployed in single-site scenarios. For more information on Db2 pureScale with HADR, refer to High availability disaster recovery (HADR) in Db2 pureScale environments.