Understanding Cloud Database Architectures: Benefits, Trade‑offs, and Design Choices

Cloud Computing Principles and Database Impact

Cloud computing relies on disposable, interchangeable machines, which directly influences the functionality of databases deployed in the cloud. The fundamental attributes are scalability and replaceability, affecting how data is stored and accessed.

Traditional Database Classifications

Traditional databases fall into two broad categories: parallel‑first (e.g., MongoDB, Teradata) and single‑system‑first (e.g., PostgreSQL, MySQL). Each design carries inherent limitations that can hinder support for certain features, especially when migrating to cloud environments.

Examples of Feature Gaps

For instance, Greenplum supports sequences while Redshift does not, despite both being PostgreSQL derivatives. Conversely, BigQuery lacks sequences, whereas Teradata provides them, illustrating that feature parity is not guaranteed across similar engines.

Cloud Database Landscape

Cloud databases belong to the same categories, with newer systems favoring parallel‑first designs. In the single‑system‑first space, cloud offerings such as Heroku PostgreSQL, Amazon Aurora, Google Cloud SQL, and Azure SQL focus on managing costs, upgrades, and reliability (RPO/RTO).

Parallel‑first cloud databases split into two sub‑categories:

SQL/relational : BigQuery, Snowflake, Redshift, Spark, Azure Synapse

DHT/NoSQL : BigTable, DynamoDB, Cassandra, Redis

These distinctions depend more on physical data layout and hash‑based key access than on the presence of an SQL‑like language.

Storage Model Constraints

Parallel‑first relational databases often rely on native cloud storage systems that expose a limited get‑object/put‑object API. This API permits partitioning but restricts high‑performance random access, limiting the implementation of advanced persistent structures such as indexes and mutable data handling.

Consequently, cloud implementations tend to use sequential reads/writes on micro‑partitions rather than traditional indexing. Even when external indexes are built, the underlying storage API may make address or byte‑offset usage difficult.

Advantages of Cloud Computing

Managed Infrastructure : Providers handle hardware, OS, security, and operational stacks, allowing users to focus on core business.

Scalability & Efficiency : Resources can be elastically allocated, enabling load smoothing, high utilization, and cost‑effective handling of large workloads.

Cost Transparency : Detailed accounting exposes storage and compute costs per query, supporting cost‑benefit analysis and optimization.

Rapid Experimentation : Free trials and easy provisioning let users start small and scale as needed.

Weaknesses and Trade‑offs

Fault Domains : Distributed cloud infrastructure introduces new failure modes; a single component failure can affect multiple services.

Feature Immaturity : Many cloud databases lack mature features found in on‑prem products; some capabilities may never be added.

Performance Overheads : Certain operations (e.g., INSERT, UPDATE, DELETE) can be less efficient due to storage model constraints.

Latency & Vendor Lock‑in : Additional network latency and dependence on a specific provider’s services complicate migrations and increase costs.

Cost Management Complexity : While scaling is easy, controlling expenses can be challenging; exceeding budget caps may cause job rejections.

Choosing a Cloud Database

When evaluating providers, identify the essential workflow attributes and verify that the chosen platform supports them, even if this requires integrating multiple complementary services. No single product offers every feature, so prioritize core requirements over marketing claims.