Entities, tags and inheritance
Excubius places every resource in exactly one node of a tree of entities, and lets you group resources across the tree with tags. Settings flow down the tree: what a node sets, everything below it inherits until a lower node or a single resource overrides it.
The entity tree
An entity is a node that represents whatever your organisation needs: a company, a client, an environment, a team. Excubius offers the kinds root, tenant, client, environment and group, but they are labels, not rules, so use as many levels as you need.
In this example, Areito Corporation is a managed service provider at the root. Each client gets its own branch, and each branch is split into environments. The two SQL Server instances live once in the tree, under the availability group they belong to.
Moving a resource or an entity to another node changes what it inherits from then on. The audit log records every move.
What is inherited
| Setting | Set it on | Overridden by |
|---|---|---|
| Thresholds of every check | Any entity | A lower entity or a single resource |
| Notification channels (Slack) | Any entity | A lower entity that has channels of its own |
| Maintenance windows | Any entity or one resource | — windows add up |
| Roles of users and directory groups | Any entity | — the highest role on the path wins |
| Default theme and logo | Any entity | A lower entity, or the user's own theme |
| Registration mode for new users | Any entity | A lower entity |
| Jira settings: project, issue type, priorities, labels | Any entity | A lower entity |
| Area weights of the security score | Any entity | A lower entity |
| Credentials | Any entity | — usable by the resources of that entity and below |
The screen that edits a value always shows where the effective value comes from: the baseline of the catalog, an entity by name, or this resource. That way nobody has to guess why one server alerts at 70% and another at 80%.
How a threshold is resolved
Excubius starts from the baseline that the check's catalog entry declares, with its source (Microsoft documentation, the CIS Benchmark, Azure Monitor recommendations, DBA community practice or your RPO and RTO). Each entity on the path to the resource can replace a value; the resource itself can replace it last. The most specific value wins. Thresholds and baselines explains the details.
Set thresholds as high in the tree as you can and override only the exceptions. A rule set once on Production covers every server added there later.
Tags
Tags are key:value labels on resources, such as env:prod, role:sql or client:ciguapa. A resource lives in one place of the tree, but tags let you find it from any angle without duplicating it.
Use tags for what cuts across the tree: every production SQL Server of every client, everything that holds personal data, everything with a linked server. Today you use them to filter the resource list with a tag rule, such as env:prod role:sql, and to group the Hex view of the Overview. Saved views that pin a tag rule are planned.
Recommended minimum tags: env, role and client. Also give each resource its Criticality, from 1 (low) to 5 (critical): it weighs the resource in the security score and puts critical resources in trouble on the Overview.
Roles follow the tree too
A role is given to a user or a directory group on a node and applies to everything below it. A person can be Entity admin for one client and Operator for another. On any node, the effective role is the highest one found on the node and its ancestors, counting the person's own roles and those of their directory groups.
| Role | Can |
|---|---|
| Viewer | Read dashboards, alerts and resource detail; change their own settings |
| Operator | Everything a viewer can, plus acknowledge, silence, assign and resolve alerts, create Jira tickets and plan maintenance windows |
| Editor | Everything an operator can, plus change the checks and thresholds of resources and accept security risks |
| Entity admin | Everything an editor can, plus manage the branch's entities, threshold defaults, resources, credentials, sources, users, roles, invitations, registration, channels, Jira settings and status page |
A Super admin role, given only on the root, covers every branch and the settings of the whole installation.
What the tree means for what you see:
- Lists only contain what lies in your branches: entities, resources, alerts, the Overview, credentials, channels and users. If your roles start at Ciguapa Inc., your entity tree starts there too.
- Anything of another branch does not exist for you. Something you can see but not change, such as an alert for a viewer, shows Your role does not allow this.
- Roles are read on every request: removing a role or deactivating a user takes effect on their next request.
- Entity admins give roles only on entities they administer, and never a role above their own.
Credentials belong to an entity
A credential, such as a monitoring login, belongs to the entity where it was created and can be used by the resources of that entity and of every entity below it. Deciding where the collector connects with it is reserved to the administrators of the credential's entity: creating a resource, or changing its connection details, parent or credential, needs the entity admin role on the entity of every credential the collector could then use. Otherwise the administrator of one client could aim a login shared from above at a host of their choosing.
Directory groups
With Microsoft Entra ID, a role can be given on an entity to a directory group, named entra: followed by the group's object ID, instead of to each person. Excubius reads a person's groups from their sign-in at every sign-in, so adding someone to the group gives them the role at their next sign-in, and removing them takes it away then. To cut access at once, deactivate the user. Directory groups are part of the Enterprise plan. Groups from Active Directory or LDAP, synchronization through Microsoft Graph for people in more than 200 groups, and groups created in Excubius are planned.
Who can join, per entity
Each entity decides how new people get an account, and the entities below it inherit the choice until one sets its own:
| Mode | What happens |
|---|---|
| By invitation | The default. An administrator sends a link; the person signs in with Microsoft or Google and gets the invited role. Nobody else gets in |
| Allowed domains | Anyone whose work address is in one of the listed domains gets an account on first sign-in, with the Default role (Viewer unless set) on the entity that defines it. Suited to internal staff |
A Google account counts for a domain only when it is a Google Workspace account of that domain; personal Google accounts always need an invitation. Registration that waits for an administrator's approval is planned.