Hi,
we are currently running Virtuoso 08.03.3335 on RHEL 9 with :
Enabled Cluster Extension
Enabled Replication Extension
We would like to implement an Elastic/Fault-Tolerant cluster using two physical RHEL 9 servers and four Virtuoso server processes:
RHEL-01: Host1 + Host3
RHEL-02: Host2 + Host4
with the following fault-tolerant host groups:
Group A: Host1 + Host2
Group B: Host3 + Host4
Could you please confirm whether our current license permits running these four Virtuoso server processes concurrently?
If not, could you clarify how many Virtuoso licenses or license entitlements are required for this topology and whether licensing is calculated per physical server, CPU/core, Virtuoso server process, or cluster?
Could you also confirm that our current license entitlement supports the combined use of the Cluster Extension and Replication Extension for this architecture?
Licence coverage for four server processes
Your current licence allows unlimited database connections, the Cluster Extension and the Replication Extension, and it can be used on both RHEL 9 servers. It’s a 16-CPU affinity licence. This doesn’t limit how many CPUs the servers can have. Instead, Virtuoso on each server is restricted to using at most 16 of that machine’s available CPUs. Licensing isn’t based on the number of Virtuoso server processes, so running two Virtuoso server processes on each server (Host1 + Host3 on RHEL-01, Host2 + Host4 on RHEL-02) is covered, and you don’t need any extra licences for this topology.
Please note that the two Virtuoso processes on each server will share that server’s 16 licensed CPUs. Take this into account when sizing the thread settings for each process (e.g. ServerThreads and ThreadsPerQuery in the [Parameters] section). If you find that 16 CPUs per server is not enough for your workload, please contact us to discuss a licence with a higher CPU limit.
Using the Cluster and Replication Extensions together
Your licence includes both extensions, and they can be used together. Note that in a fault-tolerant Elastic Cluster, the copying of data between members of a host group (Host1 ↔ Host2 and Host3 ↔ Host4) is handled by the Cluster Extension itself. You only need the Replication Extension if you also want to replicate to a separate Virtuoso instance, for example a standby or disaster-recovery copy at another site.
Comments on the proposed topology
The layout you describe is the recommended one for two physical servers. Each fault-tolerant group has one member on each machine, so if a whole server is lost, every group still has a working member and the cluster keeps serving. Some points to plan for:
- Interconnect: use a dedicated, low-latency and ideally redundant network link between RHEL-01 and RHEL-02 for cluster traffic. With only two physical machines, a network split between them is the main risk, so this link should be as resilient as possible.
- Resources per process: each Virtuoso process needs its own database directory, ports and memory settings. With two processes per server, size NumberOfBuffers and MaxDirtyBuffers so that both processes together fit in the server’s RAM, leaving headroom for the operating system. Where possible, put each process’s database files on separate disks or volumes.
- Virtuoso version: 08.03.3335 is the current Virtuoso Enterprise release, so you’re up to date. A new release is planned in the coming weeks. We’ll let you know when it’s available so you can decide whether to deploy it before going live or as a later upgrade, and we’ll be happy to advise.
- Failure testing: before go-live, test stopping a single process, a whole server and the network link, and practise bringing a failed member back into its group.