AT&T Still Cloudy On CaaS

AT&T's new cloud computing service introduced on Monday is arguably the biggest endorsement yet of an industry that's come under scrutiny for high profile failures. The latest was last month's Sidekick outage where users were taken offline after a server failure wiped out their data. Two weeks later, the Danger/Microsoft team who hosted the server were still working on restoring the data. Fall out from the service in the near term was significant. Just over a third of respondents to a survey from Strategic Technology Analytics indicated that they would review their cloud decision because of the failure.

Yet cloud computing for the enterprise continues to interest CIOs and executives. AT&T says that it was only after feedback from its User Advisory Councils, some of which comprise Fortune 100 CIOs, that it moved into the technology. Enterprises indicated that the technology has emerged from being a niche offer to one that could be used to extend existing compute resources.

The AT&T service, called Synaptic Compute as a Service (CaaS), will take square aim at such organizations. Unlike services from Amazon and IBM, AT&T will leverage it private network to provide secure connection backs to its data center. The service is also available over the Internet. Perhaps in a nod to the recent cloud fiascoes, AT&T further emphasized that the service would be monitored and managed around the clock. Service Level Agreements (SLAs) are at 99.9 percent of uptime. Organizations will be able to order their infrastructure through a Web portal.

As with Verizon's CaaS announced earlier in the year, AT&T's technology is based on VMware's VSphere virtualization platform. The service will use Sun Microsystems' cloud computing API, for example, for creating value-added services.

Three server sizes are available with the service: Small (1 CPU and 4 GB of memory); Medium (2 CPUs and 8 GB of memory); and Large (4 CPUs and 16 GB of memory). Each image will come with 100 GB of storage with up to 2 TB of additional disk storage per virtual server available, or the option to connect to AT&T's Synaptic Storage as a Service for more dynamic expansion.AT&T's entry is important insofar as the company is still a favorite when it comes to enterprise data networking. This is particularly important because while cloud services offer cost gains, they can also transform data center agility. Large organizations will be able to respond to network events by ratcheting up or lowering server bandwidth as needed. AT&T will be a logical supplier for a cloud solution.

At the same time, service providers are often challenged to explain how they differ from competitors in the same area. It's no different here. AT&T emphasis on the private network vs. Internet is an important distinction, but it's not enough. Organizations will need to answer three big questions:

1.    What are the SLAs being offered and what do they include? The existing information about 99.9 percent uptime is too vague. What exactly is covered by the SLA? Data center infrastructure? The network? When can those outages occurs? Sure, 99.9 percent sounds good, but not when it really translates into at least nine hours of downtime per year.

2.    How does the service integrate into existing data center operations? The whole idea behind a cloud datacenter service is that virtual machines can be brought up as needed. Today, organizations can already deploy a virtual server in about 18 minutes. Reducing that time even further will probably only appeal to the most demanding environments and will require a process that's been very well designed.

3.    Finally, what about greater detail on pricing? AT&T hasn't released any details on the pricing for its compute service. Until it does, all of the details around this great sounding offer will be very cloudy indeed.