Documentation / Platform glossary / Multi-tenancy
Multi-tenancy is an architecture where one running application serves many customer organizations, each in its own logical boundary. Infrastructure and code are shared; data and configuration are not. It is the model behind most enterprise SaaS, and the reason updates reach every customer at once.
Multi-tenant designs vary in how much they share, and the choice has real consequences.
Shared everything runs one application against one database, with tenant scope enforced at the data layer. It is the most efficient to operate and the most demanding to build correctly, because separation depends on enforcement being systematic rather than remembered per query.
Shared application, separate databases gives each tenant its own schema or database behind a common application. Separation is more evident, at the cost of migration and connection management complexity that grows with tenant count.
Single tenancy dedicates a deployment per customer. Isolation is strongest, and so is operational cost — every upgrade, patch, and incident is multiplied.
The trade-off that matters is not efficiency but the noisy neighbour problem: shared capacity means one tenant's load can affect others unless resources are governed. Shared infrastructure also means a shared blast radius, which is why isolation must constrain operational impact and not only data access.
The compensating benefit is uniformity. One version in production means security fixes apply everywhere immediately, rather than being scheduled per customer.
Multi-tenancy is what makes continuous delivery and consistent security posture economically possible. It is also the reason isolation questions are asked so persistently: the efficiency comes from sharing, and every shared resource is a place separation must be enforced.
Evaluating a multi-tenant platform is therefore less about the sharing model than about where and how the boundary is enforced.
QueryTek serves many organizations from shared product families while keeping each tenant's data, users, and configuration distinct. Isolation is enforced systematically rather than per feature. Deployment topology and enforcement detail are not published.
