Joining EC2 fleets to a customer's existing AD: no AD Connector, across regions
Joining an autoscaling EC2 desktop fleet to a customer's existing Active Directory, connectionless, across regions, with no AWS Managed AD or AD Connector, plus the orphan-cleanup discipline that keeps the directory tidy.
Read Article →You can't always get a GPU, so we made the desktop appear where you can
GPUs are scarce, so we place each session in whatever region has capacity, and use ONTAP + FlexCache so the user's multi-gigabyte files and Windows profile follow them there, behaving like local storage.
Read Article →Derive, don't duplicate: localized Windows AMIs without a second pipeline
Supporting another language usually means a second image pipeline to maintain forever. Instead we derive a localized image chain from the English layers at plan time: one source of truth, N languages. (Describes the design approach.)
Read Article →JIT VDI: instant GPU desktops without paying for an idle fleet
Cold-booting a GPU Windows desktop takes several minutes; a pre-joined warm pool brings one to life in about a minute. Here's how we made GPU VDI feel instant without paying for an idle fleet.
Read Article →Closing the golden-image feedback loop: AWSTOE + an AI that iterates to a goal
Baking a whole image to test one component takes the better part of an hour. We run components standalone on a warm box with AWSTOE, give each a machine-checkable verify goal, and let AI agents iterate until it passes.
Read Article →A production-grade DCV environment in hours, not weeks
How we stand up a full, multi-region Amazon DCV estate (session brokers, connection gateways, an access portal, and autoscaling GPU desktop fleets) in hours instead of weeks, and the engineering decisions behind it.
Read Article →
.png)
