As unified communications applications become more prevalent, it’s only natural that users and enterprises will come to rely on them more and more, making it mission critical that they be always available. If users can’t get to the apps that they’ve come to rely on to do their job, be it voice, instant messaging, conferencing, desktop video or what-have-you, productivity will naturally suffer.
Companies need to take steps, then, to ensure reliability of their UC implementations. And those steps will be somewhat different depending on whether the company uses a premises-based or cloud-based UC solution. In this two-part blog series, I’ll take a look at some of the best practices for each, starting with the premises-based option.
Defining Levels of UC Reliability Right For Your Company
The single word that crops up most often when you dig into pretty much any discussion of IT reliability is redundancy, and it’s no different with UC. The only real question is, how much redundancy is enough?
“We find that customers invest more in redundancy as the criticalness of the UC system increases,” says Rich Winslow, Senior Director, Product Management for ShoreTel. “A 10 person donut shop will be less concerned than a 500 user staffing agency about an outage.”
True enough. AudioCodes, which sells VoIP and UC products and services, sees four tiers when it comes to the level of resiliency customers demand with respect to UC, says Alan Percy, senior director of marketing for the company:
None: If the link is down, employees do without or use their cell phones. “A surprising number of businesses say that’s acceptable,” Percy says. Like 10-person donut shops, probably.
Resilience at the desktop handset level: This is a rather new phenomenon, where companies have a second data center outfitted with UC servers to serve as a backup. That second data center is pre-programmed into users’ handsets such that if the connection to the primary site fails, it automatically fails over to the backup site.
Branch office resilience: If the WAN link goes down, employees can still use some subset of their usual UC tools to communicate with one another on-site and outside via a backup connection, whether public Internet, PSTN, MPLS or the like.
Full-fledged backup: Full-blown UC implementations in each office, again with a backup wide-area connection.
Options for Achieving UC Redundancy
Lots of options exist for achieving your preferred level of redundancy and reliability.
One is to choose a UC architecture that is distributed in nature, such as ShoreTel’s. It puts call processing intelligence in appliances or virtual applications that run at each site. In the event of a WAN failure, users at each site can at least communicate with others at the same site. With centralized call intelligence, if the WAN fails, users at remote sites are out of luck.
Another similar tack is the Microsoft approach for its Lync users, which is to use a Survivable Branch Appliance (SBA) to protect branch offices. An SBA is essentially a slimmed down version of Lync, giving users a subset of Lync functions when the primary WAN link is unavailable, namely telephony. In the event the WAN link goes down, the SBA fails over to a backup connection – PSTN, public Internet, backup MPLS or SIP trunk - enabling Lync users to continue to at least make voice calls.
Users who need full-fledged backup in each location will require a full UC implementation at each location. For example, AudioCodes sells another Lync appliance called One Box 365. As the name implies, it is a full-fledged Lync implementation integrated with Microsoft Office 365, with survivability features similar to the SBA.
Building Unified Communications Redundancy into the Network
Of course the idea of redundancy isn’t exactly new to UC; many of the same tools and techniques that companies have used for years to build redundancy into their networks apply.
ShoreTel’s Winslow offered up a number of them:
Implement call control devices in a redundant, “N+1” configuration, so there’s a backup in case the primary fails
Have more than one call control appliance with PSTN connections spread across them to eliminate any single point of failure
Configure outbound routing to use PSTN connections from any site so if you lose local PSTN connectivity, outbound calls can grab other PSTN trunks
Use redundant uplinks from your call controller to the network
Have a redundant network fabric
Have your UC elements, including PoE switches for phones, on power backup, meaning protected by UPSs
Have two service providers providing network access to the location
Obviously some of those options – notably having two service providers – will be more expensive than others. Which means every company has to decide where their level of risk tolerance falls on the donut shop/staffing agency spectrum.