>>The bottom line, as we all know, is that IBM wants to keep the margins
up on
>>engines running traditional workloads, while competing with cheaper iron
for
>>work that they perceive might otherwise be in danger of migrating there.
>Basically says it all - even IBM admits it.
I do not speak for IBM in any official capacity, but no, IBM does not
admit any such thing. IBM is indeed a publicly traded company with a duty
to shareholders to maximize profits (or so my economics textbooks told
me), but the only (durable) way to do this is to deliver excellent
products that people want to buy in ever greater quantities. And you
can't do that in the technology business without spending serious coin
($1.2 billion on the System z9, for example; a goodly fraction of that
just on tools/utilities) and beating competition (fairly). No company can
say, "I think we'll keep our margins high this month." Doesn't work, at
least not for long.
If we just (artifically -- I'll buy your argument for half a minute here)
look only at CPs, the acquisition prices have collapsed and the operating
costs have fallen even faster. In fact, since the PCM era prices and
costs have fallen *faster*. Here's what's happened just in the past few
years (and just with "traditional" workloads), all to lower costs:
- PSLC (Parallel Sysplex License Charges)
- WLC (Workload License Charges), with more and more products individually
SCRT'ed
- the other LCs (EWLC, etc.)
- conversion of certain products (e.g. Tivoli System Automation) to OTC
(customer's discretion)
- z/OS.e
- z/VM price reductions, OTC, quantity discounts, etc.
- huge drops in memory prices (and big increases in base memory
configurations)
- softcapping, LPARs, and more LPARs
- On/Off Capacity On Demand ("processor for a day")
- z890 finer grained capacity settings and lower minimum configuration
(~26 MIPS)
- IBM entering the tools/utilities business in direct (and ethical,
professional, but very aggressive) competition with ISVs (on all
dimensions: lower price, better function, and far better terms and
conditions -- no MIPS tier, "MIPS on the floor," or capacity upgrade
pricing abuses)
- zIIPs (yes, these help some DB2 "traditional," such as utility-related
processing)
- "technology dividends" (a System z9 with the same LSPR capacity as a
z900 is rated at 19% fewer MSUs)
- much faster I/O (MIDAW, 2 Gb FICON, more channels, etc.), which ends up
lowering CPU
- continued code efficiencies (e.g. CICS 3.1 thread-safe code benefits,
the new z/OS 1.8 XML services)
- lower power, cooling, and smaller physical footprints
- lower and more competitive storage costs
- much better (and faster) crypto offload
- more push out on the "statistical tail" of (already best) planned and
unplanned outage avoidance (e.g. DB2 online schema changes, IMS online
reorg, System z9 "live" processor service replacement) (downtime = cost)
- elimination of the Sysplex Timers and their maintenance (pending 2006)
- elimination of unnecessary external assist boxes and their maintenance
(e.g. CCL replacing 374x equipment)
- more and better instruction assists (e.g. Unicode/EBCIDIC/ASCII
conversions)
- the quiet revolution in tape (faster, cheaper, better, VTS)
- OSA Express (lowering CPU for network-related processing)
And, if you cast the net only very slightly wider, those new workloads now
can enjoy "fastpath" Hipersocket and in-LPAR access to "traditional"
resources, reducing the amount of work (CPU) those traditional subsystems
have to do. (Architectural proximity has nice benefits.)
And I'm sure I forgot a whole bunch of other stuff, but that's off the top
of my head.
IBM has a business plan (and the ability to deliver) to help our mainframe
customers reduce their total costs substantially, every year. (Does
Microsoft? Does Oracle?) If we haven't proved it to you yet with the
list above, feel free to keep watching from the sidelines or from another
"stadium." The sidelines and other stadiums keep getting more and more
expensive, though.
>Leaves a lot of old customers swinging in the breeze though as Denis
>says. I don't see a lot of new workload selling mainframe MIPS here.
The independent figures -- admittedly global, so I don't know your
location -- say exactly opposite. Both "traditional" and "new workloads"
are growing quite rapidly (the latter faster, but both growing). The MIPS
graphs are pretty astonishing over the past 5+ years. Honestly, IBM is
making its money here the old fashioned way: investing huge sums (widening
the technology lead and expanding mainframe roles), getting bigger sales
volumes each year, and driving down prices, thus driving more demand;
loop, repeat.
>Parallel sysplex (and associated pricing model) didn't seem to do the
>job either - consequently we now see IBM releasing humungous boxes, and
>encouraging shops to consolidate back to single footprint.
Actually, IBM encourages consolidation to fewer footprints (not
necessarily "single"), consistent with objectives such as "penalty boxes"
(if still caught by unfair T&Cs), Parallel Sysplex, GDPS, etc. It costs
less money and works better. Most customers consider this a good thing.
>"new workload" is apparently the current mantra being drummed into the
>IBM workforce.
>Let's hope it works, else there might be more of us competing for yet
>fewer job opportunities.
I would not have a job if it weren't for "new workload." Or "traditional
workload," for that matter. Both are important. This is not an
either/or. Most mainframers consider it a very good thing when new
applications, new projects, and new work comes to the platform that works
the best and that they manage so well. In fact, the smart mainframers
actively make that happen within their organizations.
Said another way, business must go on, and it will: customers must be
served, accounts opened, loans processed, new markets explored. If you're
not helping your business accomplish such goals, what are you doing?
[ "Today's moment of mainframe boosterism is brought to you by the letter
Z and the number 9..." ]
- - - - -
Timothy F. Sipples
Consulting Enterprise Software Architect, z9/zSeries
IBM Japan, Ltd.
E-Mail: [EMAIL PROTECTED]
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html
