Total Pageviews

Thursday, September 20, 2012

Sometimes you just have to cut and paste...

There are times when you can't improve on perfection, so when that occurs the best thing to do is cut and paste!

The Technology Economics of the Mainframe: Mainframe Computing Still Growing in Banking

Despite perceptions, mainframe computing continues to grow in banking. But from an economic standpoint, that may not be a bad thing.
August 14, 2012

Contrary to conventional wisdom, mainframe computing is growing. In fact, financial services has been passed by only a few sectors in terms of growth in mainframe MIPS, or million instructions per second. But maybe this is good news: Multiyear results show that "mainframe-heavy" organizations are more economically efficient in supporting business computational demands and have more upward scalability than distributed-server-heavy organizations.

Decisions about computer platform choices and options typically are made without consideration of true business impact from a cost-of-goods or other perspective. As a consequence of what we know about technology economics today, however, platform choices can be based on factual criteria and should not be decided as a "fashion statement."

During the 50-year history of what we now call "technology economics," it has always been clear that demand for computing is increasing and that upward expense pressure is a fact of life in what many have called the Information Age. Between 2006 and 2010, demand for processing cycles (MIPS, servers and the like) slowly approached an 18% annual growth rate in the big banks, while storage demand has been growing at 45% or more per year.


With infrastructure spending -- on computing power, networks, storage, help desks and so on -- historically accounting for 57% of overall IT expense, it is likely that it is the largest component of an organization's "IT Cost of Goods." As such, it is worthy of investigation and analysis.

Previous research that exlpored the dynamics of platform economics indicated that firms with a mainframe computing platform bias ("mainframe heavy") exhibited a lower IT Cost of Goods and overall IT costs in situations in which the mainframe was a suitable platform. Conversely, "server heavy" firms were at an economic disadvantage -- higher IT Cost of Goods and overall infrastructure costs. Additional research updates these earlier findings, continuing to chart the interaction (and value) of computing choices and real bottom-line business impacts.

Key cost of goods metrics were identified for the sectors under study. Within each sector, analyses were performed to determine average levels of both mainframe and distributed server usage relative to business volumes/revenue. Within each sector, two groups were identified -- "mainframe heavy" and "distributed-server heavy," relative to average levels of usage.
Within these two groups (by industry), IT Cost of Goods was computed and compared.
mainframe computing
The research database for this study contained data from 498 companies across 20 sectors, spanning the years 2008 to 2011. Data elements include the amount of computational resources along with key business performance parameters.

Across the 498 companies studied, on average, computational needs grew far faster than revenue. MIPS capacity grew at 2.33 times the rate of revenue growth, while distributed server deployments grew at 3.5 times the rate of revenue growth. Additionally, firms that had higher mainframe growth had 25% lower distributed server growth and exhibited approximately

67 percent more-effective cost containment than those with less mainframe intensity. The implication is that the required computational growth is roughly three times more economically efficient in a mainframe environment.

Further, organizations with high mainframe intensity had 39% more upward scalability in that they could support revenue growth with 61% less investment than those that were distributed-server-intense. And organizations with high mainframe intensity maintained their leverage in terms of lower IT Cost of Goods. Across sectors, the gap widened by 3%; in banking, the gap widened by 2%, with mainframe platforms maintaining an astounding 67% cost advantage at the unit cost-per-core transaction level.
This research reveals a pattern that indicates that mainframe-heavy organizations are more economically efficient in supporting the computational demands of increased revenue than distributed-server-heavy organizations. Such patterns are critical to observe and understand as computational demand increases in the global economy, in business and government, and in our daily lives.

It is likely that 2012 is the "technology economic tipping point" -- the point at which demand for computing growth outstrips the ability of Moore's Law to offset increased costs. This perhaps is also the point at which organizations that have not learned by codifying the patterns of technology costs and value will see their tech expenses spiral out of control, leading to uninformed demands to cut IT expenses. On the flip side, those oganizations that have an understanding of technology economics will find themselves in a position of extreme competitive advantage.

Howard A. Rubin is founder of Rubin Worldwide, a research and advisory firm focused on the economics of business technology. Howard.Rubin@rubinworldwide.com

Thursday, September 13, 2012

1st Mainframedebate review

Yesterday between 4-5pm was one of the most stressful hours of my life!!!  As most of you will already know, yesterday we held the 1st ever mainframe debate on Twitter. This event came about from an idea I had a couple of months ago.  Being a Twitter nut, IBMer and also keen cyclist doesn't always overlap but on this ocassion it did, let me explain.

On Facebook I follow a number of cycling companies and one of them regularly hosts a Facebook session for an hour with one of it's mechanics where anybody can post and ask questions...you would think pretty mundane but about 40 people ask questions in the hour.  Wind forward and I am running Mark Anzani's (IBM VP for System z) diary while he is in the UK this week.  So I decide that it would be good to have Mark as the guest on a Twitter debate, playing the role of the mechanic!

Well after numerous conference calls, an IBM internal newsletter article and many swapped emails we get to yesterday at 3.30pm, thirty minutes before the debate is due to kick off.  In the room we have:

  1. Mark Anzani
  2. Jeffrey Biamonte
  3. Richard Gamblin
  4. Paul Yarrow
  5. Geoff Eggleston

On the phone we have Carly Exum and Pratin Ashketar, in Le Guardia Airport it transpires we have Paulo Carvao another IBM VP...

On my laptop I have access to not only my own twitter account but also the main IBM_System_z account.

When we get to 4pm the chaos begins, we have questions backing up and we are brainstorming how we can answer questions that would normally require a 5-minute discussion in 140 characters...for the next 60-minutes it is a blur of switching between Twitter accounts, Sametime with you boss' boss, finding links, getting answers into 140 characters when they have been built by a room of people and general panic and chaos.

We get questions from multiple IBMers including 2 IBM VP's and 2 Business Unit Directors we even get 2 customers join in.  We get representation from Jordan, India, USA, Cuba and Italy as well as bunch of guys from IBM UK during the debate.  In the hour we get over 100 interactions and this leads to multiple subsequent retweets and mentions..

And then 5pm arrives and the fastest hour of my life has gone and we are agreeing to do it again next month.  We learnt some valuable lessons while on line, so if it was less than perfect we apologise, bear with us we will be better next time...The lessons include:
  • Don't attempt a Twitter debate on your own
  • Don't have the main laptop on a dodgy WiFi network
  • Set your Sametime status to do-not-disturb
  • Get your WW colleagues involved they helped massively in the set-up
  • Don't promise an IBM VP coffee when the canteen is closed
  • Don't expect a Poughkeepsie Chip God to be succinct

See you on the 10th October (details to follow) for what will hopefully become a regular #mainframedebate

Monday, September 10, 2012

More Mainframe Twitter debate details



September 12th 2012 #MainframeDebate: zEnterprise EC12


On Wednesday, September 12th from 11-12am EDT (16:00 - 17:00 BST), @IBM_System_z and @StevenDickens3 will be hosting a Twitter debate (Steven’s blog post) to discuss the new IBM zEnterprise EC12 and how it can help businesses to do more with less. This chat features Mark Anzani, IBM VP, Portfolio and Technical strategy and CTO for System z.

On August 28th, IBM introduced the newest zEnterprise system, the zEnterprise EC12 (zEC12). The zEC12 is designed to help enterprises effectively manage their core business processes. Improvements of the zEC12 include 50% increased capacity over its predecessor and 25% performance enhancement for transaction processing.  With the zEC12, businesses can tackle the challenges of today’s competitive business environment like never before. Read System z VP Paulo Carvao’s Smarter Computing Blog post for more information on the zEnterprise EC12.

Join IBM System z experts on September 12th at 11 a.m. ET for our first #MainframeDebate to discuss zEnterprise EC12. If there is anything you’d like to discuss, tweet questions to @IBM_System_z and @StevenDickens3

This Twitter chat will feature:
Mark Anzani, IBM, Vice President, Portfolio and Technical Strategy, System z | Follow @MarkAnzani on Twitter

Mark Anzani is the CTO and VP of Portfolio and Technical Strategy for System z. He was instrumental in the development of the zEnterprise EC12. Mark is responsible for driving long term business and product technical strategy for System z. Mark has been with IBM since 1983 and has held several positions working with IBM’s enterprise systems. In 2009, Mark became the Chief Technology Officer for System z. In July, 2012, he was became the Vice President of Portfolio and Strategy for System z. 

If you have not joined a Twitter chat or debate before, here’s how they work:
  • What: A Twitter debate is an online conversation held at a pre-arranged time following a specific hashtag. Join our chat with the hashtag #MainframeDebate. You will need a Twitter ID to take part.
  • When: Wednesday, September 12, at 11 a.m. ET
  • Where: The chat can be followed on Twitter using the hashtag #MainframeChat or you can log on and access the chat on http://twebevent.com/MainframeDebate
  • Who: Anyone who is interested in joining and has a Twitter account! Just include #MainframeDebate in your tweets!
After the #MainframeDebate:
Check out the zEnterprise EC12 announcement page. Let us know your thoughts on zEC12! Leave a comment on the Smarter Computing blog below or connect with us on Facebook, Twitter or LinkedIn.

Friday, September 7, 2012

Processor Architecture - a laymans view of the religious war

Just read, a for once, unbiased review from the Register on IBM's R&D investment in chip architecture:

http://www.theregister.co.uk/2012/09/05/ibm_z12_mainframe_engine/

Definitely worth a read for the more geeky amongst us, who find micro engineering just a little bit cool.

Then I read the blog comments and came down of my geeky high, from all of the comments by people who just want to have a knock at IBM:

http://forums.theregister.co.uk/forum/1/2012/09/05/ibm_z12_mainframe_engine/

Now I can't claim to unbiased as Big Blue feeds my kids, but the level of vitriolic spewing on these blogs just shocked me...yes me a salesman of 17 years standing...How can people generate so much passion about the various pros and cons of chip architecture.  I know the guys in Fishkill and Poughkeepsie care, but I am sure they would not get 'biblical' on another approach.

A senior IBM exec said to me recently of IBM's chasing single thread clock speed that you can compare it with the airline industry and plane size Vs speed.  What is better Concorde or a 747?  One is designed to carry ~120 passengers very quickly the other is designed to carry ~500 a lot slower.  Both have there place as approaches (and yes I know Concorde is no longer in service).

Another saying I like about the mainframe is that "we worship different gods and speak in different tongues" never could this be more true than after reading the blog comments.

For all those x86 fans out there let me sum up the mainframe with a couple of anecdotes:

  • I know of a large retail bank that has 93% of it's business logic running on 4 mainframes that cost the bank 7% of it's IT budget.  The other 40,000 servers cost 93% of the IT budget to keep running!

  • I also know of a £1bn turnover retailer that runs its entire web front end on 4 IFL's (Linux engines on the mainframe).

So come on guys give IBM a break we have a very specialized market to service and we are spending hard cash to innovate...how can that be a bad thing?