Amazon Elastic MapReduce and Microsoft’s Hadoop on Azure service are getting some new competition today with the launch of a technical preview of SunGard Availability Services‘ hosted Apache Hadoop solution, called Unified Analytics Services (UAS). According to SunGard, general availability scheduled for the second half of 2012.

UAS will setup Hadoop clusters, import data and monitor and maintain the infrastructure, freeing up CIOs from the need to purchase and maintain physical infrastructure. The clusters will be hosted in SunGard’s data centers. Customers will be able to upload data to the SunGard via the Internet, a dedicated circuit, or by sending in hard drives.

SunGard Availability Services CTO Indu Kodukula says the company will also help customers with scaling Hadoop clusters as they grow. “We’re starting at pretty large entry point (50 TB of data and higher), but both the business model and the underlying architecture will be elastic,” he says.

Instead of SLAs based on uptime, Kodukula says UAS will have SLAs focused on completion time and throughput.

Importantly, SunGard will provide isolated physical compute capacity, to address regulatory and compliance issues. Unlike Amazon Web Services, SunGard Availability discloses the physical locations of its data centers (UAS will be in SunGard’s Rancho Cordova data center to start), which clears another compliance barrier.

Services Angle

According to Kodukula, the company is in the process of creating a big data analytics consulting practice as part of its SunGard Consulting Services. But for now it will rely on its partner network to provide consulting services. I see this as critical for SunGard’s service, and another way to differentiate from AWS.

As InfoChimps, another company getting into in the hosted Hadoop business, recently pointed out, many companies still need more help figuring out what data they have and how to use it than they need help building infrastructure. Services will be a key way for existing enterprises to find big data expertise.