The Safety Net That Lets You Right-Size with Confidence
SmartPulse’s whole value comes from sizing every warehouse to its real, historical workload – which only works if you can trust that a sudden spike won’t leave people staring at a spinning cursor. Burst Protector is that trust, built in: it watches every SmartPulse-managed warehouse for queueing that its scheduled size can’t absorb, and scales it up automatically, in seconds, before the spike turns into a real slowdown – then brings it back down once the spike passes. It runs automatically on every SmartPulse warehouse; there’s nothing separate to turn on.
The reason teams over-provision “just in case” is the fear of that one bad spike. Burst Protector removes the reason.
- Runs on every SmartPulse-managed warehouse by default – the protection is already there the moment SmartPulse is
- That guarantee is what makes SmartPulse’s right-sizing safe to trust in the first place – you get the full saving on quiet hours without gambling on the busy ones, with extra protection when it’s needed
- Every burst response is visible in real time – for you to monitor and control
Every action is bounded, time-boxed, and reversible – Burst Protector is built to intervene only exactly as much as the moment requires.
- Detects a burst by comparing queued queries against the warehouse’s real concurrency capacity – not a guess, an actual measured mismatch
- Scales up within seconds of a sustained burst being confirmed, preventing slowdowns
- Every burst episode is time-boxed: if a manual change is made to the warehouse mid-burst, Burst Protector steps back and defers to it, rather than fighting it
Burst Protector doesn’t decide for you how far to go during a spike – you set that once, per warehouse, and it holds to it every time.
- Cost mode: absorbs spikes within your original, pre-optimization configuration – the cost ceiling is capped there, so mid-burst never costs more than the warehouse did before SmartPulse touched it
- Performance mode: allows a brief step above that configuration during a genuine spike (adding a cluster on Enterprise and above to handle more queries in parallel, or moving up one size on Standard to speed up the queries already running, whichever lever your Snowflake edition supports), capped wherever you set the ceiling, for when speed matters more than the extra credit.
Every action is measured and attributed, not left to guesswork.
- Every action shows up in the warehouse’s event log under its own “SmartPulse Burst Protector” entry, so you can see exactly when it fired and what it did
- Real-time stats show burst activity as it happens – you can watch a spike being absorbed live, not just read about it after the fact
- Nothing about a burst response is silent or unattributed – if a warehouse scaled up outside its schedule, the log tells you why
