Microsoft Azure · Software · YoctoIT tech page

SQL & Cosmos

The managed SQL database and the planet-scale NoSQL: application data without iron administration, with HA and patching included.

FOCUS · DATABASES WITHOUT IRONPaaS means no clusters to pamper: design, costs and performance remain
Request the Free Assessment Learn more
YoctoIT material for clients and partners · Microsoft, Azure, Entra, Defender and the other products mentioned are trademarks of Microsoft Corporation.
01 · What it is

Azure SQL & Cosmos DB, made clear.

Azure SQL brings SQL Server into PaaS: built-in high availability, automatic patching, point-in-time backups, Hyperscale to grow without migrations. Cosmos DB is the distributed NoSQL: single-digit latencies, global replication, multiple APIs (SQL, Mongo). The DBA doesn't disappear: the job changes — design, tuning, costs.

99,995%
Azure SQL's zone-redundant SLA: the HA you don't build
PITR
point-in-time restore up to 35 days: human error forgiven
<10 ms
Cosmos DB reads, SLA-guaranteed, anywhere in the world
Azure SQL & Cosmos DB
OFFICIAL MICROSOFT BRANDING · SQL & COSMOS
PORTALE AZURE REALE · CREAZIONE AZURE SQL · FONTE: MICROSOFT LEARN
REAL AZURE PORTAL · AZURE SQL CREATION · SOURCE: MICROSOFT LEARN
02 · How to use it well

The things that make the difference.

The two families

ApplicazioniERPs, web, APIs
Azure SQL DB
SQL Managed Instance
Cosmos DB
pure PaaS · full compatibility · global NoSQL
HA, backup, patches includedthe work that disappears
Cost governance (DTU/vCore/RU)the work that remains
The right engine for every piece of data

The right choice

SQL DB, Managed Instance or VM? The compatibility and cost matrix decides, not fashion.

Assisted migrations

DMS and assessments: from on-prem SQL to PaaS with window-sized downtime, not weeks.

Performance tuning

Query Store, indexes and plans: PaaS doesn't forgive badly written queries.

RUs and autoscale on Cosmos

Request Units governed: planet-scale at planet-Earth prices.

03 · In depth

vCores, Hyperscale and RUs: the models

Azure SQL: vCore with serverless (auto-pause) or provisioned, Business Critical with in-memory replicas and zone redundancy, Hyperscale scaling storage to 100TB with named replicas; Managed Instance for total compatibility (SQL Agent, cross-db). Cosmos DB: RU or serverless model, 5-level consistency (bounded staleness the golden compromise), a partition key to choose with a cool head, change feed for events and autoscale following the peaks.

  • Hyperscale — distributed page storage: 100TB and restores in minutes
  • Serverless — auto-pause and per-second: the dev/test that doesn't cost while idle
  • Managed Instance — the full instance: agent, CLR and cross-database for lift&shift
  • RU & autoscale — Cosmos's currency: the model tunes it, not hope
  • 5 consistenze — from strong to eventual: bounded staleness for healthy multi-region
  • Change feed — every write as a stream: reactive projections and integrations
04 · Numbers and lifecycle

The numbers that matter.

99,995%
zone-redundant Business Critical SLA
100 TB
the Hyperscale ceiling per database
<10 ms
Cosmos p99 reads guaranteed by SLA
35 gg
maximum PITR on Azure SQL
PaaS doesn't mean no DBA: models, partitions and plans in our care — with failover groups rehearsed.
05 · Use cases

Where it really pays off.

Modernized ERPs

The legacy's database in PaaS: patches and HA are no longer your problem.

Global apps

Cosmos for users across continents: the data close to everyone.

Consolidamento istanze

Elastic pools: dozens of small databases, a single budget.

The managed database isn't administered, it's governed: design, tuning and costs within our perimeter.