|
| Network Markitecture
I have the need to rant today...
If you are not familiar with the term "Network Markitecture," then you are
a very lucky person. Perhaps you have not been exposed to the hell that is
a company driven by sales and marketing (S&M... that's pretty much what it
boils down to for an engineer). It IS painful. It's an engineer's
worst nightmare. Network Markitecture creates a network that is so painful
to plan, administer, and repair, that it easily doubles and or triples the
workload of the engineers that "implement" it.
Engineering Driven vs. S&M Driven Companies
Engineer Driven
A network (even better... a company) driven by a QUALIFIED
Engineering Department will have a solid network, easy to maintain,
difficult to implement, yet easy on provisioning. The network will
function well, have a low problem rate, and will be scalable. When a
network engineer sits down to design a network, there are three major
priorities:
- Scalability: A non-scalable network creates an engineering
headache when the load of the network becomes greater than available
capacity. It requires RE-engineering when the limit of existing
"scalability" is reached. Engineers do not like RE-engineering.
RE-engineering requires migration. Migration is a painful,
arduous, process that is ridden with problems.
- Fault-tolerance: To an engineer, fault-tolerance means being
able to sleep at night. If a portion of the network goes down, another
part of the network makes up for it. Fault-tolerance means automatic
failover. Fault-tolerance means that it can wait until the morning.
Fault-tolerance means that your sweat and blood spent designing the
network has paid off.
- Security: There is no such thing as a stable, insecure
network. Especially within the past year, security has come to the
forefront. An engineer knows that if the network is built in an insecure
manner, he or she will be going right back to rebuild the insecure
device. Security is a blanket. Another piece of the puzzle that allows
an engineer to sleep at night. Without security, the engineer goes to
sleep hoping that it won't be this night when someone breaks in.
S&M Driven
A network designed by S&M has only one priority. Not because the S&M people
are mean (although it does seem that way sometimes), but because they do
not have all of the information needed to make an informed decision.
S&M's network design priority is:
- Beauty: It looks pretty when I print out the network diagram.
A network Marketeer looks at a piece of paper and makes decisions based on
other pictures that they have seen. Some marketeers may have a limited (or
even extensive) knowledge of network protocols and still just don't get the big
picture. Marketeers generally hang out with other Marketeers and usually hang
out with the management types. Marketeers are basically internal salesmen.
They have an idea and they sell it to management. Regardless of how ridiculous
the idea is and regardless of the man-hours it will take to deploy said idea,
in a S&M driven company, decisions get made (good or bad) that are solely based
on marketability.
Both Methods are Bad
It is impossible to stay in business with a scalable, fault-tolerant,
and secure network that no one is willing to buy. It is impossible to stay
in business with a network that everyone is willing to buy, but costs too
much money to maintain. There MUST be a happy medium. A network is
comparable to a living breathing creature. The network cannot sustain
itself without customers. The customers will not stay without a decent
network.
The "build-it-and-they-will-come" mentality relates perfectly to network
engineering. In the business world, the "build-it-and-they-will-come"
mentality has put numerous companies out of business. Many companies have
spent HUNDREDS OF MILLIONS of dollars building beautiful networks,
but did not have the customer base to warrant such a network. On the same
token, many companies have built horribly designed networks to support
a number of customers that cannot be supported.
Both methods fail miserably. The engineers see the shortcomings of a
poorly planned network. The S&M types see the shortcomings of the
unsellable network. When one side makes a decision without input from the
other, the network falls into total disarray. What you end up with is the
worst of both worlds. A network that is non-scalable, insecure, difficult
to maintain, on which it is difficult to provision new customers -- and
to top it off -- unsellable.
Management vs. S&M and Engineering
It is management's job to listen to both sides. It's management's job to
referee the eternal battle between S&M and Engineering. The very definition of
management is to make decisions. Any decisions based on half of the
information is bad a decision. Engineering departments have very different
concerns from marketing departments. Engineers will ALWAYS attempt to find the
best and easiest way to defeat a problem. S&M willl almost ALWAYS find the
best way to sell a service. Most of the time, the two are in direct conflict.
Summary
If you are in management and you are reading this, chances are that an
engineer has given you the link. Remember that, this is written from an
engineer's view of the world. If you are marketing and are reading this,
chances are that if this was not sent to you directly, you are guilty of
marketeering.
Management (read this)
- Listen to your maketing department!!! They are the ones that bring
in new services and sell your company to customers!
- Listen to your engineering department!!! They are the ones who know
what it is that you are selling!
- Think long and hard about the cost vs. benefit. Engineers like to
spend money to make things work.. Marketing likes to spend money to make
things pretty. Both are important! Think about total cost of ownership
and don't forget about the time of your salaried employees.
- I have to throw this in, because it happened to me today... DO NOT
LET S&M throw away time that an engineer has spent. If an engineer has
spent the time to build something, chances are that it is important to
engineering. Marketing doesn't see the big picture, so the importance
(to them) is minimal. NETWORK MONITORING and NETWORK METRICS are
generally not seen as revenue generating services. This is true. These
two services are considered by Engineering to be the MOST
important services of the network. The individual customer is not of
concern. The customer base as whole is of larger concern.
Engineering (read this)
- Listen to your marketing department!!! If they say something isn't
going to sell (or make a difference in the sale), think twice about what
you want to do. Marketing generally thinks about individual customers.
Engineers generally think about the entire network. Marketing doesn't
think about a customer being "down." Engineering thinks about hundreds
or thousands of customers being "down." It is YOUR job as an
engineer to make marketing see that a network wide outage causes
individual customers to go "down." I truly believe they don't see it
that way.
Marketing (read this)
- Listen to your engineering department!!! I guarantee they don't see
things the same way that you do. Remember that the decisions that you
make affect the LIVES of the engineers. For the most part, you
make decisions and rarely think about them again. The engineers think
about these decisions you've made every day (and some nights when they
are working instead of spending time with their significant others).
|
