From the desk of: Sam C. Chan
Reasons I've been against in-place*

OS Version** Upgrade since 1990s

First published: March 6, 2017
Last Updated: Feburary 2019
Addendum: Oct 30, 2019
  • empirical analysis: UNACCEPTABLE results
    • 20% no immediately discernible issues [but...]
    • 75% minor glitches requiring various remedies
    • 5% catastrophic show-stoppers
    • 0% truly perfect
  • strategically: 
    • does not represent a cost saving: easier upfront - unreliable, risky and costly in the long run.
    • even minumal steps to avert common, clear and present dangers, would nearly eat up all time savings, relative to fresh install.
    • once post-upgrade "event" occurs, the efforts & costs for subsequent remedy could be orders-of-magnitude beyond those of proper fresh install in the first place.
  • technical analysis:
    • inherently unsound approach: unlike upgrading applications, where app installer can count on stable OS foundation to perform its tasks in 1 pass... OS version upgrade is literally pulling the rug from under itself, resorting to extremely risky tricks and numerous re-re-re-re-restart, with lots of KNOWN unanticipated scenarios.
    • many applications have unusual (often, out-right inappropriate) ways of implementing things, and must be installed after OS is in-place, allowing it to modify it as it sees fit. Unexpected subsequent remodification/undoing of the app's mod to the OS, will result in unstable and unpredictable conditions.
Why do people do it then?
  1. blissful ignorance (no particular reasons, just don't know any better)
  2. once aware of problems, they still accept the risks, after deliberation:
    • incapable of performing from-scratch install
    • instead, use press-a-button approach, hope for the best
  3. mistaken believe that upgrade is cheaper:
    • FALSE - Full retail package $300 (hardly anybody would ever pay that). Upgrade $165. OEM (discounted full-only, no upgrade, for sys builders) $150. All MSRP here.
  4. obscure compatibility glitches, must resort to "carry-over" for ancient peripherals. e.g.:
    • 5-yr old Brother personal printer installed on Windows 7, EOL 3 yrs later. No updated driver for installing on Windows 8 was issued. Printer will likely (85%) remain functional after in-place upgrade of OS. 
    • need ancient (and known insecure) DotNet Framework 1.1 already installed on w7 and working, upgrade to w10, and it likely will still work. Whereas fresh install onto w10 is not supported (deprecated).
  5. system is relatively new, but extreme low-end specs. Vendor provides zero support for fresh install (no drivers). Intend to be sold and used as-in, or discard and buy new. Even some mid-end Ultrabooks are like that.
*In-place Upgrade, using the existing version OS to install the next version, overwriting itself in the process, and hopefully arrive at the other side still intact. Contrast that to Wipe & Clean Install Upgrade + reinstall of applications. Data must be backed up first in both cases.

**Edition Upgrade, on the other hand, is perfectly fine. Unlike Version upgrade, it is not changing the underlying software architecture. It's merely paying and get a new code, unlocking previously restricted features, or increasing parameters. The same software is already there all along. In-place upgrade to the same version, is also fine (safe, but not ideal), and should be referred to as in-place reinstall (aka Repair Install) / refresh (preserving settings) / reset (destructive).


SEE ALSO

Copyright @2011-2019   Bravo Technology Center  *  Bravo:GO  *  Contact Us