October 30, 2025
Fork it like it’s hot
Replacing EBS and Rethinking Postgres Storage from First Principles
New ‘Fluid Storage’ vows instant database clones; fans cheer, skeptics say cloud speeds still lag
TLDR: Tiger Cloud’s Fluid Storage promises instant database cloning and elastic scaling, now powering all free-tier Postgres. The crowd is split between excitement for agent-friendly speed and doubts about cloud performance vs. hardware, with EBS pain and “is this new or borrowed?” comparisons fueling the drama.
Tiger Cloud just dropped “Fluid Storage,” a reimagined storage layer for databases like Postgres, and the comments lit up. It claims instant clones, no downtime, and elastic scaling—and it already runs every free-tier database. The pitch: “agents are the new developers.” Cue eye-rolls, hype, and memes.
Fans like thr0w cheered: agents spinning up test environments need speed. Skeptics like the8472 zoomed in on raw numbers, arguing cloud disks still trail real hardware, citing AWS versus a gaming SSD. Meanwhile, 0xbadcafebee delivered a cathartic rant about Amazon’s EBS (Elastic Block Store): slow restores, limited volume attachments, and resizing that takes hours. The “EBS therapy group” jokes wrote themselves.
One camp thinks this finally fixes the “wait all day to copy a database” nightmare; another asks if Postgres will actually feel faster or if this is just fancier plumbing. Comparisons flew to https://xata.io/blog/xata-postgres-with-data-branching-and-pii-anonymization and https://www.simplyblock.io/, with “did Tiger invent this, or remix it?” whispers.
For newcomers: Postgres is a popular database; NVMe is fast solid-state storage; IOPS is a speed score. Bottom line: fork it like it’s hot is the vibe, but the speed-vs-cost debate rages on.
Key Points
- •Tiger Cloud launched Fluid Storage, a distributed block storage layer with zero-copy forks, true elasticity, and synchronous replication.
- •Fluid Storage is fully compatible with PostgreSQL and other databases/file systems, presenting as a local disk via a user-space driver.
- •Architecture includes an NVMe-backed block store, a proxy layer for copy-on-write volumes, and a user-space storage device driver.
- •Benchmarks claim a single volume can sustain 110,000+ IOPS and 1.4 GB/s throughput while preserving its elasticity and COW guarantees.
- •The article critiques Amazon EBS’s elasticity limits (resize cooldowns, inability to shrink, allocation-based billing) and says Fluid Storage now powers all Tiger Cloud free-tier databases.