6 ms·

shameless self-promotion Three ex-Pebblers went on to found Memfault(YC W19, https://memfault.com https://memfault.com), taking what they learned about managin

shameless self-promotion Three ex-Pebblers went on to found Memfault(YC W19, https://memfault.com https://memfault.com), taking what they learned about managing a fleet of IOT devices and building a suite of monitoring and OTA tools to support other hardware companies. Answering questions like "did fleet battery life get worse with yesterday's update?" get really really hard when you start managing thousands of devices. Memfault is all about making it easier to create the next Pebble. We're hiring! https://www.workatastartup.com/companies/memfault https://www.workatastartup.com/companies/memfault

RandallBrown 4y ago
> Answering questions like "did fleet battery life get worse with yesterday's update?" get really really hard when you start managing thousands of devices. To me that only seems like a question you could answer by managing thousands of devices. Not that it would be easy, just that you couldn't really answer that very accurately with only a handful of devices.
- tyhoff 4y ago
  It’s primarily about the number of hours. At Pebble, we’d be able to get a pretty good gauge on battery life with about 10-20 devices and about 24 hours worth of data. We recorded the battery’s percentage drop over the course of each hour, averaged all samples together, and then used that to estimate how long the battery charge would last. This strategy is covered here: https://interrupt.memfault.com/blog/device-heartbeat-metrics#battery-life-trends https://interrupt.memfault.com/blog/device-heartbeat-metrics... Of course it’s better with thousands of devices, but we were always surprised how accurate it was (as long as we had a few Android users in the office - it reliably got worse battery life due to worse Bluetooth performance)