Practice vocabulary for discussing isolation models with engineering teams: pool vs silo trade-offs, blast radius, noisy neighbour mitigation, and cost vs isolation language.
0 / 5 completed
1 / 5
An architect says: "We chose the pool model because the trade-off between cost and isolation made sense for our customer tier." What is the 'pool model' in multi-tenant architecture?
The pool model (also called the shared model) means tenants share infrastructure. All tenant data lives in the same database, separated by tenant_id. It is cost-efficient — one database serves thousands of tenants — but isolation is entirely logical rather than physical. The trade-off: lower cost, lower isolation. Contrast with the silo model (one database per tenant) which costs more but provides stronger isolation.
2 / 5
During an architecture review, a senior engineer says: "The blast radius of a misconfigured query in the pool model is much larger than in the silo model." What does 'blast radius' mean here?
Blast radius is a risk scoping term: if this thing fails, how big is the damage? In a pool model, a bad database migration, a resource-exhausting query, or a security bug can hit every tenant simultaneously — high blast radius. In a silo model, a failure in one tenant's database affects only that tenant — contained blast radius. Teams explicitly discuss blast radius when choosing or justifying isolation strategies.
3 / 5
A team lead explains: "We implemented noisy neighbour mitigation on our database tier." What problem does 'noisy neighbour' describe?
The noisy neighbour problem is a classic multi-tenancy challenge: in shared infrastructure, one tenant's traffic spike consumes resources that other tenants depend on. Mitigation strategies include: per-tenant rate limiting, connection pooling limits per tenant, query timeouts, read replica routing for heavy analytical tenants, and in extreme cases moving high-usage tenants to dedicated infrastructure (the 'silo' treatment).
4 / 5
What does an engineer mean when they say: "The trade-off between cost and isolation is the central architectural decision in multi-tenancy"?
The cost-isolation trade-off is the core design tension. Silo model (database per tenant): maximum isolation, highest cost, simplest security model. Pool model (shared database): lowest cost, weakest isolation, requires careful application-layer enforcement. Bridge models (schema-per-tenant) sit in between. Teams choose based on: compliance requirements (HIPAA, SOC2), customer tier (enterprise customers often demand silo), and scale economics.
5 / 5
An engineer proposes: "For our enterprise tier, we should move to a silo model and give each tenant their own database." What is the primary argument for this approach?
The silo model's value proposition for enterprise tenants: physical isolation satisfies strict compliance requirements (GDPR data residency, HIPAA PHI isolation), simplifies audit scope, eliminates noisy-neighbour risk, and gives the tenant confidence their data cannot be accessed by other tenants even through application bugs. The cost: each new enterprise tenant adds real infrastructure cost. Many SaaS platforms use a tiered model: pool for SMB, silo for enterprise.