Re:Ask HN: Where to find raspberry Pi zero or alternative un...
by bigiain on Monday July 24, 2017 @08:44PM
(#14844110)
While I understand the Pi Foundations "one per order/customer" rule - I share your frustration. There's a bunch of ideas I've got where 10 or so Pi Zeros would be both useful and affordable @ $5 ea or even $10 for the ZeroW, but I just can't buy them like that...
I wanted to make a "real working" diagram of our standard AWS platform as a wall chart - with 3 ELB load balancers, 5 autoscaling ec2 instances (3 "active" and two "spares"), and 3 "multi-az" RDS db servers - each represented by a Pi Zero, with ws2811 led strip running between them representing the network which lights up animating packet/data flow. It'd have big red killswitches next to everything, so you can push buttons to kill off bits of infrasructure and visualise how the platform responds (with the spare ec2 instances autoscaling in to replace dead ones, ELB and RDS traffic auto-rerouting). And I'd use this to run our "standard" backend, so people could connect with their phone (with a browser or test app) and "see" their own network traffic and watch how it still works even if you kill any 2 and up to 8 different parts in the right combination.
I think that'd be a really useful way to demonstrate to non-technical stakeholders why if they want better response times that "we'll get to that Monday morning" when something breaks after 5pm on a Friday - I'm going to charge them _way_ more to support their site if it's running on a $5 or $10 per month VPS than if they spend $120/month or so on AWS to host it.
(Oh, and I second zapt02's recommendation for the NextThing CHIP - I've got half a dozen of those, and they're working out really well in other projects - but I've also got another 5 on order and have been waiting several month for em...)
I wanted to make a "real working" diagram of our standard AWS platform as a wall chart - with 3 ELB load balancers, 5 autoscaling ec2 instances (3 "active" and two "spares"), and 3 "multi-az" RDS db servers - each represented by a Pi Zero, with ws2811 led strip running between them representing the network which lights up animating packet/data flow. It'd have big red killswitches next to everything, so you can push buttons to kill off bits of infrasructure and visualise how the platform responds (with the spare ec2 instances autoscaling in to replace dead ones, ELB and RDS traffic auto-rerouting). And I'd use this to run our "standard" backend, so people could connect with their phone (with a browser or test app) and "see" their own network traffic and watch how it still works even if you kill any 2 and up to 8 different parts in the right combination.
I think that'd be a really useful way to demonstrate to non-technical stakeholders why if they want better response times that "we'll get to that Monday morning" when something breaks after 5pm on a Friday - I'm going to charge them _way_ more to support their site if it's running on a $5 or $10 per month VPS than if they spend $120/month or so on AWS to host it.
(Oh, and I second zapt02's recommendation for the NextThing CHIP - I've got half a dozen of those, and they're working out really well in other projects - but I've also got another 5 on order and have been waiting several month for em...)