6. Platform as a service (PaaS)
Whether you're going to the public cloud or deploying your own private cloud, hand-installing each operating system, application server, and application, then selecting the number of servers or VMs you'll deploy it to in advance is so last decade. PaaS is the way to go. Scale just-in-time and automate away repetitive tasks. Platforms we've worked with extensively include CloudFoundry,CloudBees, and OpenShift.
7. Cloud IDEs
Recently we were cleaning out a closet and my nine-year-old didn't know what was the big, metal box thing he couldn't lift. "What is it?" we asked him from the other room. "I don't know, it's some Dell thing." We realized he genuinely didn't know what a tower computer was since I've used laptops for his entire life (though he would've recognized a 1U server from seeing them at my work for Hadoop testing). I think cloud IDEs offer promise that the next generation won't know what a laptop is. Why install an IDE on a hard disk? Why not just open a browser and start coding with, say, Codenvy or Cloud9, especially if you're deploying to a PaaS anyhow?
Whether you use MapReduce for complex analytics or simply want to make log analysis and audit logs a fire-and-forget-oh-just-write-it-quickly event, Hadoop is a monster that's set the industry on fire. If you haven't at least put out a pilot project, this is the year to do so. If you've already dipped your toe in, I expect you'll jump in fully this coming year.
9. Clustered/distributed file systems
From Gluster to HDFS, scalable storage is the thing. This is the year your SAN gets a rethink. At the very least, it's time for a pilot if you haven't done one already. I predict a lot of hybrid approaches in the meantime.
Sign up for Computerworld eNewsletters.