rubenf
rubenf9mo ago

hosted or self-hosted ?

hosted or self-hosted ?
13 Replies
Unknown User
Unknown User9mo ago
Message Not Public
Sign In & Join Server To View
rubenf
rubenf9mo ago
10$/operator (1/2 seat) 20$/developer (1 seats) Then you pay per number of vCPUs used for your workers so that it is linear to the infrastructure cost rather than number of workflows $50/vCPU e.g, 5 developpers, 2 workers with 1vCPU each = 520 + 250 = $200/mo if you are a small enough company, you might be able to fit the pro version (less than 10 FTEs) We provide dedicated supports among others
Unknown User
Unknown User9mo ago
Message Not Public
Sign In & Join Server To View
rubenf
rubenf9mo ago
We charge based on usage, even though you host them (it's a pricing different than number of executions, and is easier to predict on your side)
Unknown User
Unknown User9mo ago
Message Not Public
Sign In & Join Server To View
rubenf
rubenf9mo ago
Windmill is considerably cheaper than any of the alternatives, without any fixed cost and it grows linearly with your usage. The best proxy we found for your usage is vCPU which is easy to understand and predict since it depends only on the infrastructure you use.
Unknown User
Unknown User9mo ago
Message Not Public
Sign In & Join Server To View
rubenf
rubenf9mo ago
you can use a single setup for all your environments. And in theory yes it does count but if you're not doing much compute on your staging instance and it's just to mimic prod perfectly, we are happy to remove them from the count The private npm route is not fully supported without EE, and we will have to post updates soon about which cases are supported and which cases aren't
Unknown User
Unknown User9mo ago
Message Not Public
Sign In & Join Server To View
rubenf
rubenf9mo ago
you only need to restore the database for it to be fully operational again and windmill is designed to never crash, never need restart. If the machines get interrupted for any reason, it will handle it well. Better if it is a SIGTERM than a full machine interruption but in both case the system will still be operational and resume the jobs that got interrupted
Unknown User
Unknown User9mo ago
Message Not Public
Sign In & Join Server To View
rubenf
rubenf9mo ago
it's safe but to be extra safe for major upgrades, we recommend scaling workers to 0 and then back to their numbers. it's a queue of jobs so not jobs will be lost and all workers will wait to finish their jobs before scaling down
Unknown User
Unknown User9mo ago
Message Not Public
Sign In & Join Server To View