January 21, 2012 |
In a hurry? Skip to
conclusion/recommendations at bottom |
In short: Not recommended! It is certainly not to be installed in a haphazard manner, as it is potentially disruptive
in a typical production system, at this juncture:
- hardware resource requirements
- SP3 code is significantly bulkier, thus consumes more memory& disk space
- during upgrade process, 2G of hard disk space is required for temp
files, etc.
- after installation, it takes up 1.2G ~ 1.8G of additional disk
space, half of which is due by backup version of system
files, the remainder belong to uninstalling mechanism, and newly
expanded executables.
- system partition would grow by 1.2G ~ 1.8G, making GHOSTing more
difficult
- compatibility issues
- very old XP systems typically harbor numerous minor corruption,
which likely would prevent the upgrade script from patching up
executables to the new version
- some applications have no SP3 compatible version, other would
support fresh install onto SP3 platform, but not survive an in-place
upgrade.
- some device drivers have no SP3 support, or work only as a fresh
install, but cannot be migrated over to SP3 by Microsoft's patching
- logistical process considerations
- during the upgrade process, remote IT will lose connection
- upon first reboot, Microsoft resets many system settings to default, prompt
corresponding reaction from IT
- upon first reboot, Windows built-in firewall would be activated, preventing all remote access, requiring
local intervention, via verbal instructions.
- all these mean it'd have to be performed during business hours, with
all the implied downtime and distraction to staff
- risk assessment based on my empirical data
- typical scenario:
- 6- to 10-year old system
- 256M to 512M memory
- 30G to 120G hard drive
- lots of legacy applications, often from defunct vendors
- some legacy peripherals (scanner/printer/wifi), without
up-to-date driver availability, due to EOL policy
- 25% risk of minor glitches, mostly sidestep-able, or possible to
live with them
- 10% risk of major glitches, enough to be considered operationally
unacceptable
- 2% risk of catastrophic melt-down: runaway costs + time drain +
stress
- necessity and justifications
- when vendors proclaim SP3 is "required," 90% of the time, it
reflects only their arbitrary administrative policy, not a technical decision/assessment
- it is entirely understandable and fair, for vendors to demand
certain baselines, so as to reduce their workload & risks.
- in most cases, such "requirements" can be skipped,
and the program would work, and it is commonplace practice by IT
dept. everywhere, per their internal priorities and risks assessment.
- majority of the time, the real reason things don't work lie
elsewhere, and it has to be addressed even after SP3 upgrade is
performed.
- of course, it's possible for some software vendor to implement a
check, and lock out program functions, or refuse to install, upon
detecting non-compliance. In which case, you must make a decision.
- strategic decisions/directions/alternatives
- we definitely will not casually upgrade all XP stations at once,
at minimum: start with 1 pilot station, allow reasonable wait
period to evaluate & validate, then consider rolling out to
remainder
- it is oft acceptable to forgo certain apps which prompted such
upgrade: entirely or at selected stations
- it might be sensible to designate a single
station (or a few)
for such apps
- consider virtualization for off-loading such apps
- consider fresh install of XP SP3
from ground up
- more drastic alternatives, such as: pushing up
hardware replacement schedule
- Bottomline: Avg. $40~$75/station in best case scenario, when
nothing went wrong, just spending time taking care of all the basic
tasks. $150+/station is not uncommon; and in nightmare scenarios, the
sky is the limit.
|