November 28, 2007, 5:17 PM — Just when you’re thinking about taking some well-deserved time off work
to enjoy the holidays with your family, you realize with frustration that you
won’t be able to put your Blackberry down for even a minute. Your company
had a big promotion throughout December and your Web site came crashing down
because it couldn’t handle all the business. Too bad the marketing department
didn’t let the IT department know about this Web promotion in advance.
To make matters worse, all those extra temporary employees that were brought
in to handle new business are sitting idle because they can’t access your
system to help customers make their last-minute purchases. And, as if that weren’t
enough, someone on your team decided to do an unplanned system update at the
worst possible time.
Here are some practical tips for preventing these holiday misadventures.
They shop 'til you gridlock
You’ve really got a problem when marketing announces a new promotion just
before the holidays and you don’t have the capacity to run it. In the past,
IT simply went out and bought more servers, but that option is too expensive
and impractical in today’s complex environment. It’s like installing
a second fireplace because Santa’s too big to get into your house, when
what he really needs is a wider chimney.
The best approach isn’t to buy new boxes; it’s to do a better job
of planning and capacity management. All too often, though, people don’t
know how to do effective capacity planning, and they're wary of putting virtualization
on a production machine. So, they end up with a room full of machines that gobble
up electricity. They forget to assess their baseline. They don’t do an
inventory to determine what boxes are on the floor, what applications are running
on those boxes, what capacity they are currently using, and what service level
agreements they have.
The good news is there’s software that can help you determine what's running
on each of your servers and then analyze whether you can run virtualization
or share workloads. This kind of information can help you know if it’s
possible to double the transaction load without killing the response time. A
completed “what-if analysis” could help you predict what’s needed
to keep Web site traffic running smoothly during the holidays. With all this
information in hand, you could double the workload or combine two servers and
see the impact on performance. It’s so much more effective than just crossing
your fingers and hoping everything works. And don’t discount the importance
of keeping customers happy so they’ll return to your site again and again.
The clock is ticking, but the temps can’t work
Just as Santa needs to get his elves up and running for the big day, a lot of
retailers must get their temporary staff set up to meet the holiday demand.
Within a few hours of HR telling you the people are coming, they could be walking
through the door. If you have to get them set up manually, it can take far too
long before they’re ready to work.
There's no point in bringing people on board right before the holiday and then
having them sit idle for a week because they don’t have what they need
to do the job they were hired for. They need a laptop, IDs, passwords, and software.
And after they leave you’ll need to roll that all back again.
All of this requires a series of critical steps to get your temporary employees
up and running. Configuring a server manually can take as much as 80 hours because
of all the tasks that typically tend to get done in sequence. The first person
does his or her job, passes it down to the next person, who does that job, then
passes it on to the next person, and so on. In a worst-case scenario, this can
go on for weeks. Yet, with the right identity management and service automation
solutions to automate and trigger all of the necessary actions, you can get
someone on board in less than an hour.
Someone makes the right change at the worst possible time
You may be just ready for the holiday rush, and all of a sudden your systems
aren’t working properly. All you know is that a Web server is down. It
could be running out of capacity. It could be an error. Or it could be that
someone is applying a change or a patch without thinking about the ramifications.
Maybe someone scheduled the Web server to go offline for an upgrade without
realizing what the machine does or its value to the business. There go hundreds
of thousands of dollars of business headed for someone else’s Web site.
Ouch!
