6.4 In Summary 535
more appropriate for dealing with the overall timeline of a performance
6.4 In Summary
Performance monitoring and system tuning is not very far from being an
art, and it would be very presumptuous of me to say that I am a master of
this particular art. Great artistic work, in my opinion, is typically a combi-
nation of talent (often a gift) and know-how. I hope that with the informa-
tion provided here, you have enough know-how to check and monitor your
Exchange messaging network performance, and I trust you will have the tal-
ent to interpret and ﬁne-tune the data. I cannot, unfortunately, transmit
any kind of talent, but I hope to provide you with best practices that can be
helpful as you do your day-to-day job.
Among the key things to remember, I would like to mention the follow-
ing: do not hesitate to equip your messaging infrastructure with vertical and
complete monitoring solutions. They will let you sleep peacefully and take
care of properly interfacing a complex messaging environment with opera-
tional support staff by providing events and associated actions that can help
the system run smoothly and address problems before the users are even
aware of them (this is a nice goal to have for complex and distributed sys-
tem monitoring). Vertical solutions aim at drilling down a particular appli-
cation and providing an intelligent way to interpret the data and report
Using a moving
average to a
536 6.4 In Summary
events and activity. Complete monitoring solutions provide a 30,000-foot
view of the computing infrastructure and federate the many operating sys-
tems and application environments under a single viewpoint of reporting,
often required in large-scale computing environments that require 24 × 7
monitoring and typically a follow-the-sun operating model.
Be critical of whatever counter you monitor and how you interpret the
data. You will ﬁnd hundreds of performance counters on a Windows 2003
Server running Exchange 2003, but with just one or two counters, you can
quickly determine whether your machine is falling behind or providing a
good service level, from user response time and message throughput and
background activities perspectives.
With this, I wish you a great performance monitoring of your Exchange
servers. You will ﬁnd that this activity can be extremely rewarding both per-
sonally and to your organization if you can draw conclusions that help in
reducing your IT infrastructure costs and provide a better service to your
user communities. Remember that you should strive for a balanced infra-
structure, especially regarding your servers and storage conﬁgurations, and,
where you can, use acceptable budgets to satisfy to a certain service level.
Throwing CPU and RAM into a box do not help if your storage compo-
nent is falling behind.
Knowing that a server is not performing adequately is ﬁne, and it is very
important to address before your users complain. Knowing why the server is
not performing up to your standards is better. Knowing how to improve
your server performance is the best approach.
Without a sound analysis and comprehension of Microsoft Exchange
2003 performance and monitoring, you don’t stand a chance of achieving
these three points.