# Monday, March 16, 2009

Most of you know about Uber Bandwidth's growing reputation for its high performance, high value network.  But even the best network becomes of no value if power stops and servers grind to a screeching halt.  That's a tender subject if a power outage at another data center has cost your business money.  You may recall how Amazon's retail operation lost an estimated $31,000 per minute during a virtually complete shutdown on June 6, 2008 that lasted nearly 5 hours (do the math—a $9.3 million loss).  At the Uber Center, we take great strides to ensure that won't happen to our customers, resulting in 100% uptime since opening this facility nearly 3 years ago.  From the redundant dual bus electrical infrastructure all the way down to fully managed branch circuit monitoring, the Uber Center's high standards assures the performance of every amp delivered to the rack.

At every potential point of failure, we try to exceed industry best practices in order to exceed industry reliability.  Here are a few examples for why our power to the server has experienced zero down time.  First, data centers sometimes offer cut-rate pricing by installing used equipment for power, cooling, etc.  That can be "penny wise and pound foolish" since older equipment typically incorporates outdated technology, lower efficiency and is closer to its MTBF (mean time between failures).  The Uber Center's power (and all other) equipment is all brand new, best-of-breed (such as Mitsubishi UPSs) and remains under full maintenance contracts with 4h onsite service SLAs to guarantee reliable delivery of power.

Second, the majority of data centers use "standby rated" generators because they are less expensive per kilowatt.  When they are new, well-maintained and from a reputable manufacturer, they generally work fine as long as outages do not extend for days or weeks at a time.  But data centers must be prepared for the unexpected.  That's why the Uber Center uses "prime rated" generators in accordance with ISO8528 that are designed to run continuously for up to 365 days per year without a strain.  Prime rated generators have maximum power available at a variable load for an unlimited number of hours, and they generally include a 10% overload capability for short durations.  That could make the difference between uptime and downtime for your servers!

Third, most data centers "exercise" (test) their generators at least biweekly or monthly to be sure they will start when needed.  But then they run the generators for a relatively short duration in order to save diesel fuel.  After all, these thirsty beasts can drink up to 75 gallons of diesel per hour for every megawatt of power.  And at today's volatile fuel prices, that's a lot of money just for testing!  But the short-run strategy of many data centers is false economy.  Why?  A diesel generator needs to run for at least an hour at a time to properly lubricate engine components, prevent carbon and contaminant build up, heat up generator windings and expel excess moisture.  Furthermore, even with fuel additives and proper storage, diesel has a limited "tank life" and should be completely refreshed with replacement fuel every 24-36 months or less.  In order to achieve these parameters and ensure superior generator reliability, the Uber Center exercises its generators biweekly and for 1-2 hours per exercise at full load in real-time tests using a 50+ item checklist.  Yes, it costs us more time and money, but it provides our customers with 100% uptime.  And that's what the Uber Center is known for.

Monday, March 16, 2009 10:07:45 AM (Eastern Standard Time, UTC-05:00)  #    -
Datacenter
# Wednesday, February 25, 2009

You're right, our flagship data center is not in or near a major city.  And for good reason:  lower security risk and lower operating cost, both of which are cited by our customers as major reasons why they colocate here.  In fact, it was customer demand at our six other data centers—all in major metro areas—that drove our decision to build the Uber Center in Asheville, North Carolina.  Major cities are not only higher in building and operating costs (passed on to customers) but, perhaps more importantly, they are generally much higher risk for both natural (hurricane, tornado, earthquake, flood, blizzard) and man-made disasters (terrorism, riots, crime, industrial hazards, etc.).  Yet in choosing an ultra-secure, non-urban location, we also insisted on good accessibility (not the middle of the Sahara) and great connectivity.  After many months of research, we concluded the best location in the U.S. to achieve these goals was Asheville, North Carolina, a mountain shielded plateau less than a day's drive for over half (some say 75%) of the U.S. population and now VERY well connected to numerous Tier 1 Internet providers.  Our customers enjoy lower risk and lower cost without sacrificing good accessibility and great connectivity.  We even provide hours of free remote-hands-and-eyes by certified support engineers to enhance your accessibility to servers.  So is our flagship data center really no-where?  Actually, it's now-here :)  Still have doubts?  Check out our white paper on "Choosing a Datacenter Location."

Wednesday, February 25, 2009 11:25:17 PM (Eastern Standard Time, UTC-05:00)  #    -
Datacenter
Archive
<September 2010>
SunMonTueWedThuFriSat
2930311234
567891011
12131415161718
19202122232425
262728293012
3456789
About the author/Disclaimer

Disclaimer
The opinions expressed herein are my own personal opinions and do not represent my employer's view in any way.

© Copyright 2010
Netriplex LLC

All Content © 2010, Netriplex LLC