Recently at the PASS (http://www.sqlpass.org) Community Summit
in Denver this year, I had a very interesting conversation with Andy Warren (http://www.endtoendtraining.com/)
on a number of topics primarily designed to get my blood pressure as high as
possible.
Some of these talks actually degrade into intelligent discourses like
security, disaster recovery and other fun SQL Server topics. I've given lots of
talks over high availability and disaster recovery best practices and personal
experiences, and over the years I have focused more and more at the higher end
of SQL Server deployments where down time are to be avoided at all cost.
When
Andy and I started talking about clustering and other technologies we started comparing
notes on past experiences. I started hammering the book on randomly testing
backups, why clusters aren’t the end all be all to uptime when he said
something that startled me for a second. “We didn’t test backups and we didn’t
have a need for a cluster, when I talked to the boss and showed him the cost
for the cluster it was more than a day’s worth of downtime the boss didn’t
think we needed it and the business could afford the loss if it came to that.”
Wow,
one, the boss knew what the cost of doing business was for a day, and two
Andy was able to communicate the technical aspects in a way the boss could
understand. Sometimes, No down time is just the nature of doing business in a
particular field. But, for the majority of SMB's it's just not needed.
I have
come up against the CTO/CIO who believes that they MUST have 100% up-time,
especially after 9/11, without understanding what or why they are truly asking
for or what that will cost. Some believe that if it cost say 15,000 dollars to
get to 95% up-time that getting to 100% may cost say 2 times or as much as 5
times that much. Next thing you know an implementation is in full swing.
Sometimes, poorly planned and poorly understood. The front line guys may know
what is really in store but the people at the top simply push on.
Eventually,
they call in outside help. Once I tell them they could have a onetime cost as
much as a 100 times the original "estimate" of $15,000 plus the reoccurring
cost to sustain infrastructure and staff for 100% up-time the top starts to
really re-think what 100% uptime buys them. Time and again I always try to
promote truly knowing the cost of your business. Sometimes that's helping the
top brass understand what NOT to invest in, what the true cost are if they do,
and what really is an acceptable loss to the business.