It's hard to turn on the TV, read the newspaper or talk to your neighbor without hearing about the impact of high gasoline prices. However, I believe that much of the attention paid to skyrocketing prices for gasoline has more to do with a lack of information historical context that skews our perceptions. For example, the most recent issue of Fast Company magazine calculated that the 1908 price of gasoline at 18 cents per gallon is equivalent to $3.90 per gallon today in the United States, adjusted for inflation. That means that gasoline is not overpriced today, but instead has simply caught up with the price that Americans paid 100 years ago.
I think a similar situation exists with regard to the concerns expressed over the recent outages at Google in their online application infrastructure, as well as concerns about the reliability of hosted e-mail and other applications in general. Some decision makers will point to these outages as justification for keeping all of their critical applications in-house. They might point to well-publicized reports about hosted provider’s outages and believe that hosted equals lower reliability when, in fact, a hosted offering might be more reliable than their own infrastructure, despite the high profile outages. In short, the attention paid to hosted providers’ outages, coupled with the lack of information on outages in on-premise systems, leads many to believe that hosted providers’ outages are worse than they really are.
In comparing the reliability of hosted applications with that of the in-house infrastructure that they might replace, it’s important to ask yourself some tough questions. For example, are hosted applications less reliable than your on-premise infrastructure? How much downtime has your e-mail or unified communications infrastructure experienced during the past year? How many IT staff-hours were required to resolve each downtime incident? How impervious is your e-mail and application infrastructure to major disasters like hurricanes, or minor disasters like daylong power outages? I’m not arguing that all e-mail systems, unified communication systems, applications and other parts of the IT infrastructure should be migrated to a hosted model. There are solutions that can make the on-premise infrastructure extraordinarily reliable and redundant, and there are advantages to keeping things in-house in many cases. However, I am arguing in favor of doing the research to provide an apples-to-apples comparison of hosted vs. on-premise offerings so that decisions can be made as accurately as possible. It’s important that information – not perceptions – shape our decision-making. |
