Implementing SSL for a limited-resource platform is not an easy task, given that the protocol itself was originally designed for maximum security, not optimal performance (it was likely assumed that SSL implementations would be able to use the relatively vast resources of PCs). However, SSL makes perfect sense for many embedded applications, especially for HTML-based interfaces and remote configuration and monitoring. Despite the large resource requirements of the protocol, it is flexible enough to provide us with many possibilities for saving those resources. This paper covers the design decisions faced by an embedded SSL implementor and how those decisions affect resource utilization and adherence to the protocol.