Redis is the proven category leader for in-memory caching and real-time data . fast, widely deployed, and highly rated . but the 2024-2025 license saga, recurring critical CVEs, and costs that spike at scale make it workload- and risk-dependent.
Low-latency caching and session storage for high-traffic apps (users report 'absolutely awesome' caching performance on massive analytics workloads)
Serverless apps with spiky, unpredictable traffic
Who it's not for
Cost-sensitive teams expecting flat costs as data grows . Redis Cloud users name cost 'the biggest downside as you scale,' and cheaper per-GiB managed options (Google Memorystore, Upstash) and Dragonfly (claims up to 80% savings, same API) exist
Steady high-throughput workloads . per-request costs pile up
Privacy
Active and serious CVE history in 2025-2026, headlined by the critical RediShell RCE (CVE-2025-49844), a use-after-free bug affecting all versions released over 13 years (~60,000 exposed servers; no confirmed in-the-wild exploitation at disclosure).10
No public breaches; SOC 2 and audit logs offered; one 2022 CVE in the Upstash auth adapter.
Public sentiment
Ratings are strong: roughly 4.5/5 across ~170-180 G2 reviews and 4.7/5 (212 ratings) on Gartner Peer Insights.⁴
Developers rate it 5.0/5 on Product Hunt for serverless simplicity, but Trustpilot sits at a middling 3.7.
Biggest gotcha
Pricing trap: costs balloon as you scale . G2 users explicitly call cost 'the biggest downside of Redis Cloud as you scale'; benchmark against per-GiB providers and Dragonfly's claimed 80% savings before committing²
Kafka is being deprecated . don't build new pipelines on it.
Pick Redis when
Low-latency caching and session storage for high-traffic apps (users report 'absolutely awesome' caching performance on massive analytics workloads)
Teams wanting a fully managed, multi-cloud service (Redis Cloud integrates with Google Cloud and Azure)
Organizations that value a mature product with strong peer ratings (4.5/5 on G2, 4.7/5 on Gartner Peer Insights)
Performance-critical in-memory workloads where 'always-on' operation with minimal launch overhead matters
When Redis is not a fit
Cost-sensitive teams expecting flat costs as data grows . Redis Cloud users name cost 'the biggest downside as you scale,' and cheaper per-GiB managed options (Google Memorystore, Upstash) and Dragonfly (claims up to 80% savings, same API) exist
Organizations requiring unambiguous OSI open-source licensing . Redis dropped its open-source license in 2024, triggering the Valkey fork, and only re-opened in 2025; compliance teams should not assume stability
Anyone planning to use it as a primary durable relational database . practitioners warn it 'is wonderful' for certain use cases but is frequently misused, with published guides on the ways to shoot yourself in the foot with it
Teams without the ops discipline to harden and patch . defaults assume trusted networks, and critical CVEs like RediShell required prompt upgrades across 13 years of affected releases
Pick Upstash when
Serverless apps with spiky, unpredictable traffic
Indie devs and startups on usage-based pricing
AI apps needing vector search plus caching
Vercel/Next.js projects
When Upstash is not a fit
Steady high-throughput workloads . per-request costs pile up
