I know this question has been asked many times before but in the
enterprise space it seems that ITIL is always a given. My thoughts are
maybe changes in the IT industry might change the need for ITIL. To be
clear on this I am not proposing to know the answer to this question.
Some Industry Considerations
Five years ago when I would teach a class to enterprise customers I
would always ask all the customers how many servers do you manage? On a
good day I would get maybe 10k and only a couple of super
infrastructure banks would answer above 20k. Recently I have been
attending a number of open source meetings and I am meeting people who
tell me that their infrastructures have over 100k servers and I finding
more and more that these numbers are actually small in the new WEB 2.0
world. I have talked to some consultants who are working with RackSpace
and they have told me they have over 100k servers. A recent article by
Robin Harris suggests that Google might have over a million cores. I
can’t image how many servers Amazon is farming with their S3/EC2
offerings.
Changing the Game
Companies like Facebook are adding 350k user’s a day and doubling
ever six months. Virtual World networks like Second Life and Kaneva are
growing with numbers that are mind blowing. How do companies cope with
change in these types of environments? My belief is that they have two
options. One, they go and blow and don’t focus on traditional ITM/ESM
techniques or two they change the game. If you look at what Google is
doing in Portland on the Columbia River, they are indeed changing the
game. They are using free software and cheap hardware to build what
they call “Power Provisioning for a Warehouse-sized Computer”. I
recently attended an IBM session where their development lab in Markham
Ontario is completely virtual and is run by a giant provisioning system
called Tornado. IBM developers in this facility select their
provisioned system from a self service portal and its all on demand.
Perhaps the best example of a company changing the game is Amazon’s.
Amazon processes over 4 million purchase transactions per day utilizing
over 150 different network services to deliver these successful
transactions. They are building their infrastructure based on the
Google Three Rules.
Cheap Servers - Mass produced low end servers, free software. Energy efficient switches and unmanaged switches.
Expect Failure - Cheap hardware will fail. Therefore build recovery into the software.
Scalability - Create monster scalability. Google has created clusters that exceed 8000.
Enterprises like Google and Amazon are defining a new type of data
center. I compare it to RAID 1 for servers. Plan for failures the way
large companies plan for redundant electricity. This kind of future
could normalize IT to the status of electricity that is a true
commodity. If you take a closer look at Amazon’s Elastic Computing
Cloud (EC2), today you are guaranteed 99.999% availability for your
servers and your total investment is a light switch and service bill
with no ITIL required.
John Willis
johnmwillis.com
