October 05, 2010
ITEXPO: U.K. Service Provider Hears Concerns on Hosted IP Communications Services
By Gary Kim, Contributing Editor
An attendee from the United Kingdom asked a question at my ITEXPO "SIP Trunking" panel that I hadn't heard for some time, but is a basic question every salesperson sooner, rather than later, will be asked by prospects.
"How do you answer objections about the reliability of a hosted IP telephony service, compared to an existing telephone service, when a customer asks about the new dependency on the broadband access connection?" the conference attendee said.
That's a good question. No device or network is completely "reliable," meaning there are "no" outages. The generally-accepted rules about specifying reliability in the computing and communications are fairly standard.
The "gold standard" has been 99.999 percent availability, meaning 5.26 minutes of downtime per year. Keep in mind the stringency: The "five nines" standard means the entire network, end to end, including failure rates of every active device and every application, cumulatively fails 5.26 minutes a year.
Also keep in mind the other cruel reality here: The 99.999 percent availability means much more than that each discrete device and application works at that rate. It means the entire end-to-end service works that well. That is crucial because the end-to-end availability means multiplying the "less than 1.0" value of each discrete device and application by the discrete values of all the other elements.
As you recall, multiplying a series of numbers, each with a value less than "1," will give you a smaller value each time one additional integer is added to the calculation. For a highly-complex system, that necessarily means redundancy, since the actual reliability of most devices actually does not approach such levels. In the computing industry, or even for carrier data services, availability is quoted at about 98 percent availability. That means redundancy is essential.
Consider that 98 percent availability means 7.30 days of outage every year, about 14.4 hours every month or 3.36 hours every week. And that is for every discrete computing application or appliance. There is no way a network or platform can reach 98 percent availability if it only has a couple of active components, let alone an entire trans-continental or global value chain.
Software systems, in fact, require "planned outages." Updates have to be loaded, for example. So redundant or parallel systems are required to cope with the outages an administrator will deliberately cause.
So 99.99 percent availability equates to 52.56 minutes of downtime per year; 99.9 percent availability equates to 8.76 hours of downtime per year; 99.5 percent availability equates to 43.8 hours downtime per year; while 99 percent availability equates to 87.5 hours downtime per year.
In any sales discussion with a potential IP telephony buyer, it is reasonable to reframe the discussion about the reliability of IP telephony, compared to current phone or data service, by facing the availability question directly.
There is, for example, no point in worrying too narrowly about circuit failure when there are many other sources of failure, each of which can disrupt communications, and which require remediation.
Loss of electrical power, hardware failure or loss of telecom service collectively accounted for about 82 percent of the outages experienced by some 200 medium and large businesses over roughly the last year, in the U.S. market, a CDW (News - Alert) has found. The figures for electrical system reliability might well be different in the United Kingdom, though.
Power loss ranked as the top cause of business disruptions over the past year, with one-third of businesses reporting it prompted their most recent disruption. Hardware failures caused 29 percent of network outages, followed by a loss of telecom services to facilities (21 percent).
That statistic alone provides one of the "talking points." To the extent that power failure will disable all the phones, computers and servers in an office or facility, nearly a third of all communication system failures can be protected against by using standby power at a site. Few small businesses do so.
Premises hardware failure seems to account for another 29 percent of outages. So the argument might include the observation that adding a standby power system and avoiding use of a premises switch could, in principle, eliminate up to 53 percent of all outages.
The generally "unmovable" fact is that about 21 percent of failures are caused by a loss of the communications circuit, so backup is needed for that part of the service, no matter what fixed-line access technology is used. And one way of backing up service is to use a hosted IP telephony system that can automatically re-route calls to organization mobile phones and at least some wireless-connected PCs (so that administration functions are available even if a landline connection is cut).
Re-framing the original question helps answer the question. There are many sources of outage more important that the communication circuit, no matter which type is in question. But hardening the electrical system cures up to a third of outages, while IP telephony itself, plus mobile backup, will harden the circuit part of the value chain.
In other words, the issue is not that IP telephony is "less reliable" than conventional telephony, but that all access circuits will face some amount of outage. In fact, IP telephony is part of the solution to such outages, when used with protection of at least some premises systems and devices, plus the overlay of alternate paths (wireless).
While 82 percent of the 200 businesses completing the survey felt confident that their IT resources could sustain disruptions and support operations effectively, 97 percent admitted network disruptions had detrimental effects on their businesses in the last year.
Also, about 1,800 smaller businesses reported network disruption of four hours or more within the last year. CDW estimates that such network outages cost U.S. businesses $1.7 billion in lost profits last year.
"The three most common causes of IT outages are addressable," said Norm Lillis, CDW vice president, system solutions.
Gary Kim is a contributing editor for TMCnet. To read more of Gary’s articles, please visit his columnist page.
Edited by Tammy Wolf