The DB jungle guide: "How to select the right database"

This list is compiled from 2 years of NoSQL consulting and has been presented on many conferences (video here), articles (e.g. here) and in the worlds first NoSQL Books (in german).

Cluster 1: Know & Segment your data

Analyze & Categorize it:

• Domain-Data
• Log-Data
• Event-Data
• Message-Data
• critical Data
• Business-Data
• Meta-Data
• temp Data
• Session-Data
• Geo Data
• etc.

Data- / Storage Model:

• relational
• column-o
• doc-alike
• graphs
• objects
• multivalue
• objects=ORM
• JSON
• BLOBS
• etc. (beyond bit-bucket)

Data / Type constraints:

• Data-Navigation?
• Data Amount?
• Data Komplexity (Deep XML?)
• Schema flexibility?
• Schema support needed?

Persistence design:

(Reference: (C) highscalability link to be inserted)

• Durability? On power failure?
• Memtable/SSTable; Apend-only B-tree; B-tree; On-disk linked lists; In-memory replicated; In-memory snapshots; In-memory only; Hash; Pluggable.

Cluster 2: Consistency Model

Global consistency model:

• ACID / BASE / WATER?
• Ability to (fine) tune the consistency model

CAP tradeoff:

• CP, AP, CA or tunable?

Cluster 3: Performance Dimensions

• Latency / Request behaviour / distribution [High = 10, Low = 0]
• Throughput [High = 10, Low = 0]
• High Concurrency?

Cluster 4: Query Requirements

• Typical queries look like?
• SQL needed? LINQ needed?
• BI / Analytic-Tools needed? (M/R sufficient?)
• Ad-Hoc Queries needed?
• Map/Reduce needed? Background data analytics?
• Secondary Indices
• Range queries
• Weird aggregations
• ColumnDB needed for Analytics?
• Views

Cluster 5: Architecture and Patterns

Architecture looks like:

• local, parallel, distributed / grid, service, p2p, …
• Target Platform: Hosted? Cloud? Local? Datacenter? Smartphone? Desktop?

Data Access Patterns

• read / write distribution?
• random / sequential access?
• Access Design Patterns

Cluster 6: Non functional Requirements

• Replication needed? = Rubustness
• Automatic load balancing, partitioning, and repartitioning?
• Auto-Scaling
• Text search integration? Lucene / Solr?
• Refactoring Frequency?
• 24/7 System? Live add and remove?
• Developer Qualification
• DB simplicity? (installation, configuration, development, deployment, upgrade)
• Company restrictions?
• DB diversity (allowed?)
• Security? (authentication, authorization, validation?)
• Licence Model?
• Vendor trustworthiness?
• Community support?
• Documentation?
• Company and DB dev in the future?

Costs:

• DB-Support? (responsiveness, SLA)
• Costs in general, Scaling Costs
• Sysadmin costs
• Operational Costs: (noOps)
• Safety / Backup & Restore
• Crash Resistance, Disaster Management
• Monitoring