Platform · Provisioning & Automation

Build the database. Then never onboard it.

Most platforms begin once an instance already exists, and onboarding it is your problem. SchemaPulse provisions the instance and registers it in the same run, so monitoring, alerting and backup policy apply from the moment it boots.

schemapulse — deployment managerexample data
  1. 1EngineMySQL 8.4 LTS
  2. 2TopologyGalera · 3 nodes
  3. 3HostsGCP · nam-ne2 · n2-standard-8
  4. 4ConfigurationOS tuning · THP off
  5. 5Preflight9 checks passed
  6. 6Deployrunning

Preflight

PassSSH reachability
PassCredentials
PassTarget OS supported
PassDisk capacity
PassPorts available
PassTime sync

Progress

Done02_install
Done12_os_tuning
Running20_cluster
Queued30_agent
Queued40_register

On completion the nodes register into Inventory and monitoring begins — no separate onboarding step.

Fig. 1The Deployment Manager. Engine and topology chosen, hosts targeted, preflight passed, and the Ansible roles executing in order — with OS tuning ahead of installation so the engine starts correctly the first time.

The workflow

Nine steps, in the order you meet them

  1. 01

    Select the database technology

    Engine and version, from the set SchemaPulse supports. The choice determines everything downstream — which backup engines apply, which topologies are offered, what tuning is appropriate.

  2. 02

    Select the topology

    Standalone, replication cluster, Galera, InnoDB Cluster, replica set or sharded cluster. Not a node count with a label attached — the topology drives node roles, quorum and the monitoring model that follows.

  3. 03

    Define nodes and hosts

    Provision new machines from the cloud catalog, or target hosts you already run. Sizing, region and disk are chosen here rather than discovered later.

  4. 04

    Configure the deployment

    Engine parameters, storage layout and OS tuning. Credentials come from the SSH key registry, which is keyed by environment and cluster type so staging keys cannot reach production hosts.

  5. 05

    Validate prerequisites

    A preflight catalogue runs before anything is created: SSH reachability, credentials, supported OS, disk capacity, port availability, time sync. A failed preflight costs a minute; a half-built cluster costs an afternoon.

  6. 06

    Execute

    Terraform provisions infrastructure, Ansible configures it. OS tuning — transparent huge pages, ulimits, sysctls — is applied before the engine first starts, so it boots correctly rather than being corrected afterwards.

  7. 07

    Track progress

    Per-role status as the run proceeds, with failures surfaced against the step that produced them instead of buried in a log.

  8. 08

    Register the estate

    Completed nodes enter Inventory with their versions, configuration and topology roles recorded. The agent is installed as part of the build.

  9. 09

    Begin operations

    Monitoring, alert rules, backup policy and recommendations apply from the moment the instance exists. There is no window in which a new database is running and invisible.

Coverage

What can be provisioned

  • MySQL / MariaDBStandalone · Replication Cluster · Galera Cluster · InnoDB Cluster
  • MongoDBStandalone · Replica Set · Sharded Cluster
  • PostgreSQLStandalone · Streaming Replication · Logical Replication
  • Microsoft SQL ServerStandalone · Always On Availability Groups · Failover Cluster Instances

Underneath

The automation, briefly

Terraform handles infrastructure and holds its own state; Ansible handles configuration through ordered roles. Both are driven from the control plane rather than from someone's laptop, which is the difference between a repeatable build and a build that worked once.

The ordering matters more than the tooling. OS tuning runs as a rolebefore installation, so transparent huge pages, ulimits and sysctls are correct the first time the engine starts — not applied afterwards to a process that has already made its allocation decisions.

That is as much implementation detail as this page should carry. If you want the full picture, the Architecturepage has the real deployment shape.

Why it matters

Where deployment connects back

Provisioning is what separates an operations platform from a monitoring dashboard. Every item below is configured by the build rather than by someone remembering to configure it afterwards.

Fleet & Topology
The topology chosen at step 2 is the topology SchemaPulse monitors — roles, quorum and election tracking are configured, not inferred.
Monitoring
The agent installed at step 8 begins reporting engine and host telemetry immediately.
Backup & Recovery
Backup engine capability is detected from what was actually installed, so policy options match the deployment.
Incidents & Alerting
Alert rules apply at the scope the new instance belongs to, without anyone remembering to add it.
Inventory
Versions and configuration are recorded at build time, which is what makes later drift detection meaningful.
Intelligence
Deployments are events on the same timeline as metrics and incidents — which is how "what changed before this" can answer with a schema change.