|
Dear CIO/IT managers,
This
CPTTM CIO newsletter is to bring useful news to you, CIO/IT managers in
Macau, for references without obligations, so that you can do your jobs
easier and better! Hope you like it. if you'd like to unsubscribe or
recommend your friends to subscribe, just email me at kent@cpttm.org.mo. Old
issues
are available here.
Topics in this issue:
Case study 9 on
applying ITIL at CPTTM
CPTTM has started ITIL capacity
management for a year (see here).
Here is a lesson that I'd like to share with you: You should
use different types of charts for different data. For example:
- For hard
disk space monitoring, it's enough to take a sample once a day or
once every several days. To display it, it is good to use a line
chart spanning over at least several months or one year. This
will show the speed of disk space consumption so that you know
when the space will be used up.

- Memory is not the same hard disk
space. It will not decrease with time (unless new applications are
deployed or new users are added). Therefore, it's meaningless to show
the data across several months or a year. In contrast, you can use a
line chart spanning a month (the last month) to see if there is an
overall trend of decreasing in free memory. If there is,
you need to check why. Usually this is because of new applications or
new users. In that case, if there is less than 10% of memory free, you
may want to add more memory. Note: Windows servers usually will keep
consuming memory and release it all in a cycle, making it difficult to
see if they are really in need of more memory or not.

- CPU is like memory: It will not change
over time. However, because CPU utilization can fluctuate in seconds,
it is difficult to see its overall trend in a line chart. Therefore, we
use statistics, e.g., the average utilization in the last month,
percentage of time when the utilization is over 500% (meaning 5
processes are waiting for the CPU), percentage of time when the
utilization is between 300% and 500%, between 100% and 300%,
between 50% and 100% and less than 50%. Our experience shows
that if there is 5% of more time when the utilization is over 500%, it
is already very busy. Of course, this depends on the expectation of the
users. If they feel that it's slow, we can make a proposal to the
management (spend how much in order to bring 5% to 0%). It's
up to the management to implement it or not.

No-risk
migration to ODF
ODF is the ISO approved office
document format. Because the format is open, it can be ensured that
archived documents can still be opened in the future (Can you still
open Word Star files now?). However, MS Office didn't support
ODF. The good news is, Sun
Microsystems has released a free MS Office plug-in,
enabling MS
Office to open and save files in ODF. This way, your organization
can use ODF internally while you can still receive DOC files, while
requiring no re-training of staff. What if you need to send files
to others? You can use PDF. If they need to modify the files, you can
save the ODF file as a DOC file (99% of the time this should
be fine, but you should still check if the formatting remains
unchanged) or simply send the ODF file to them and ask them to
download OpenOffice or the
plugin above.
Why
German schools can use open source software but Macau can't?
Facing the problem of whether to upgrade
to MS Office
2007 or not, 36% of all German high schools decided to upgrade,
while 25%
decided to switch to OpenOffice. At the same time, 40%
of German college students will use Linux and related software. I
can't help thinking: Why can German schools actively use open source
software but Macau can't? Is it because German students are more
independent, risk taking, wanting to learn new things and respecting
intellectual properties, while Macau students are followers, satisfied,
wanting to enjoy lifes and having no concept of intellectual
properties? It's strange that in fact open source can not only save
money and deliver security and reliability, but also improve the
quality of the people in a community. If we hope our next generation to
have more entrepreneurs, professionals, scholars instead of just casion
dealers or clerks, our schools and families should really increase the
use of open source software.
Preventing
IT project failures
The US Wisconsin State conducted an audit
on various IT projects in its various agencies to analyze the projects
with signiciant failures (some projects were delayed 6 times and costs
were doubled). The projects analyzed includes: vehicle registration
system, sales & use tax system, unemployment insurance system,
statewide IT consolidation project (sharing servers and
email, accounting, human resources, budgetting, payroll and
purchasing systems). This is a good reference for those working on
e-government. The reasons for the failures should be quite well-known
to you all: Underestimating the complexity, cost getting out of control
and schedule delayed. The report makes quite some suggestions. Here are
the few that I remember:
- Identify large scaled, high risks
projects and concentrate efforts to manage them.
- If there are challenges arising in a
project, there should be a mechanism to rescue it or terminate it.
- There must be a complete
specification before software development is started.
I personally agree to 1) and 2) very
much, but I don't think 3) can be done. The specification should be as
complete as possible, but it is just impossible to make it 100%
complete. It is even more impossible to make it 100% correct. It
is also impossible to make sure it won't change with time. Thinking
that it is possible to have a perfect specification is just burying
your head in the sand. In addition, the more detailed the specification
is,
the more errors it will contain (not specifying something means no
mistake). The more time you take to write the specification, the more
changes will have occurred. Therefore, it is not true that it should
as detailed as possible. Instead, specifying those important
things in details and leave the details unspecified for those
unimportant things.
How to handle omissions, errors and
changes? Personally I'd suggest:
- A project should be done in phases.
Each phase should be completed in half a year or 9 months (in contrast,
all the above projects in Wisconsin were spanning across several
years). What can be changed in half a year is limited.
- Get the user representatives to check
the running software frequently (e.g., weekly). If there are omissions
or errors, they can be corrected immediately.
- Allocate a buffer in the budget.
The size of the buffer depends on the scale and risk of the project. It
is common to have a buffer of 20% to 100% (coincidently, the costs of
the above projects mostly increased by 100%).
If you're interested, why not take a look
at that report.
Upcoming courses
for CIO/IT manager
There is no course for CIO/IT managers
for the moment, but there are some courses your
kids may enjoy:
Feedbacks
Any
questions, ideas or experiences to share? Contact me at
28781313 or kent@cpttm.org.mo. We also
have two other newsletters: Network
administrator newsletter and Software
developer newsletter, your staff may like to subscribe.
Until
next time,
Kent
Tong
|