I'm rather fond of PiCloud as a platform - it's good to see example projects out there using it.
Using PiCloud for implementing crawler requires some hacks, so I decided to create this as a separate package.
They're also very responsive to questions, and their package is far, far easier to use for the things I use it for than trying to build something myself via AWS, Rackspace, etc.
Disclosure: They also gave me a $500 research grant.