This article originally appeared in the January issue of INTERNET TELEPHONY magazine.

You can’t go too long shuffling through e-mails these days without seeing something on how IT is looking to move everything to the cloud – whether it’s private, public or hybrid, IaaS, PaaS or SaaS (News - Alert). But is the cloud ready for everything?

It’s true that cloud services offer significant advantages by allowing various components to run on available servers, reallocating resources as needed. With tools like VMotion or LiveMigration, these components can be moved with minimal disruption to adjust dynamically for new or different loads. However, the experience of that disruption can be dramatically different. People are used to the varying responsiveness of business applications or web pages that can take a few seconds to completely display. By contrast, even the smallest real-time media delay can feel like a walkie-talkie conversation.

Streaming media compensates for this effect by buffering content before playback starts, allowing for momentary disruption of the download to be completely hidden. Depending on bandwidth, media can sometimes be sent faster than the playback to catch up the buffering model. Even live streaming presentations are usually delayed by a couple seconds without a noticeable disruption to allow for buffering. If you consider Internet radio, for example, unless you’re listening to the presentation on a traditional radio at the same time, how would you even know?

Of course real-time media can’t be pre-buffered because it doesn’t exist. For VoIP, small jitter buffers on the receiver allow for trivial transmission variations, but these are almost never more than 90 milliseconds and quite often are variable, adjusting the buffer size based on the variations detected. The best systems relay information back to the transmitter so it can adjust packet sizes, compression techniques or other characteristics to correct for transmission anomalies.

So are clouds ready for the demands of real-time media? In some cases, no. Cloud vendors readily admit their deployments don’t have the consistent high-level responsiveness to handle real-time media constraints. It’s only in the last couple of years that UC and PBX (News - Alert) vendors have even allowed virtualized deployments, which are critical for cloud resource management. And even then, they often put additional criteria on the sizing and instances required. Reading the fine print, most indicate that moving the virtual machines is fine – as long as there are no calls running through during that process.

Despite conditions and restrictions on real-time media in the cloud, several deployment models are finding a way forward. Specialized services like Microsoft’s Office 365 and Skype (News - Alert) are available, but these clouds are dedicated to media handling and are not for use with general applications. PSTN carriers are starting to provide cloud-based services with service level agreements that allow the level of reliability needed for corporate communications. The use of hybrid clouds, where real-time media processing components are on premises and the other UC features are cloud based, is another alternative – mostly with specialized media servers, which are controlled remotely but send media directly between the endpoints in the call. Private clouds, with the appropriate sizing and restrictions, will certainly handle real-time communication needs as long as deployment guidelines are followed carefully.

That leaves general purpose public clouds. What is their fate? At the rate of cloud technology evolution, my guess is that public clouds are not far behind. We’ll see real-time media readiness sooner rather than later.

Mike Sheridan is executive vice president of worldwide sales with Aspect (News - Alert) (www.aspect.com).