RE: Help me Justify iSeries -- MIDRANGE-L

1. I would say examine the hardware config to make sure it's adequate
for DR but not much else.  For instance, in our DR environment we use
fewer disk arms so performance may be somewhat degraded but still
acceptable.  On a small 520, I'd try for just a pair of mirrored disks
(140 or 300GB or whatever) and go 10K vs. 15K RPM.  Consider the RAM
carefully.  Maybe buy a level of CPU that is 1 step down from your
production machine.  It depends on your goals for the DR box but in many
cases the company can live with degraded performance short-term.  

One thought is the DR machine will be adequate to limp along during a
'normal' DR event. If the DR outage will be extended to more than a few
weeks, the firm may simply have to pony up for extra capacity.  An extra
GB or 2 of RAM & 2 more disk drives aren't particularly expensive but if
the cost can be deferred until they're actually needed the company may
save $ since the odds of using the DR box are generally pretty slim.
Disks & RAM are also usually on the shelf so availability is good &
fast.  If you decide to push this as an option, make sure your
management understands that in an extended DR situation you may need the
$.  That you're saving them $ now but they may need to spend it later.

In our old environment, the DR box was down enough on CPU to be in a
lower software tier, which saved on SW maintenance.  We had a P40 730
for production & a P20 830 for DR.  The 830 had a little over half the
RAM.  Disk was about the same, though, due to a quirk of how everything
was purchased.  Anyway, the 830 was adequate but not swift as a DR
solution.

2. Compare the software licensing and maintenance costs.  Be sure to
include every component of each scenario: OS, app, web server, database,
backup software, network monitoring, antivirus, etc.  Add to that the
hardware maitnenance costs for the servers, maybe additional network
gear, KVM, tape drive, etc.

3. Compare the costs of all necessary hardware.  If you can run
everything on the i5 but need 3 Windows boxes (web, app, database),
compare not only the hardware costs but support costs.  You'll probably
also want a KVM switch for the Windows side.

4. Realize the additional complexities in the environment:
- Extra resources are needed on the Windows side: more switch ports,
KVM, more rack space, maybe more power & cooling, etc.
- With multiple servers web - app - db communications are limited to LAN
speed; in the iSeries that delay is as close to gone as it can get.

5. Number of Vendors in addition to the app vendor.  On iSeries, there's
1 vendor to call with a problem and they actually seem to provide good
support.  With Oracle on Windows, you have Oracle, Microsoft, your
backup vendor, etc. on the software side.  For hardware, there's your
server vendor, KVM vendor, network gear vendor, tape drive vendor, etc.

6. Compare the costs and relative effort of hardening the environment
against attacks.

7. Compare the ongoing administration costs.  This not only includes
people time, but time necessary for evaluating & applying updates, costs
associated with downtime (planned & unplanned), and updating ancillary
software like AV, backup, etc.  Add some time for hardware maintenance;
what's the impact when (not if) a hard drive fails?

Try to create a TCO for the life of the app. or 5 years, whichever is
shorter. The iSeries will probably 'lose' against Windows on some of the
comparisons, like upfront cost, but it should win the war in TCO for the
life of the app.

And try for real apples to apples comparisons.  Any attempts to stack
the deck in the iSeries' favor will be obvious and will greatly diminish
the odds of being taken seriously.

For us, the cost of administering the Windows environment is so high
that the iSeries is easily a cheaper platform in the long run.