As a follow-up to the recent "Another BIG Mainframe Bites the Dust" thread -- and apropos to a couple of other ongoing threads -- I received kind permission from Mr. Sangho Yoon to post on this listserv the following email he sent to me the other day. There is a lot to be ruminated upon in his note. Jon <snip> Hi there,
I am the Sangho Yoon from Samsung Life Insurance in Korea that has been repeatedly misquoted through numerous press releases regarding our mainframe rehosting project. I was killing some time at LAX, on my way back to Seoul and I stumbled across this thread of messages regarding our project and I read it with a mix of interest, amusement, and horror. First of all, we did not embark on this project purely from a cost perspective. Of the three main criteria for our success - reliability, performance, and cost - the financial aspect was the least important. If, through our 18 months of benchmark testing of the various solutions available for rehosting, we felt that the reliability and performance was not viable, this project would have died a quiet death. Secondly, I was somewhat amused by comments regarding whether we will survive or not. Samsung Life insurance is the largest insurer in Korea, with annual revenues in excess of US$25 billion and assets of over US$100 billion. We rank among the top 15-20 financial institutions in the world. Rumors of our demise are greatly exaggerated. It would be difficult to address every comment I've read in this thread but suffice it to say that we did not embark on this journey recklessly. Although the migration project itself only took 12 months, we spent the better part of a year evaluating and testing various alternatives before we decided that this was the right way to go for us. I will attempt to address some of the most common questions. We have now been running on our new 3x64 way HP Integrity servers in production for 107 days with zero downtime. FYI, our production environment is made up of a 3 node Oracle 10g RAC cluster and 3 application servers in a load balancing configuration. We use EMC DMX 3000 series storage arrays in a SAN that is for the most part dedicated to this system. Our online and batch performance has actually gotten better than what we were experiencing on the IBM mainframes. Our old system had 7000 MIPS and during peak times, we'd be pushing 100% CPU utilization. We rarely exceed 60% utilization today. Samsung Life has over 40 million active policies that are processed by this system. 35,000 independent insurance agents create millions of transactions daily. We've successfully completed month end and quarter end processing on the new system - in shorter time that it used to take us. We're very confident about our cost savings figures and in fact, they are conservative if anything. We've literally unplugged the mainframes and in fact have sold them - they are not coming back. It was an intense period of 12 months that got us to where we are today. I have lots of grey hairs now that weren't there a year ago. I make no claims that this type of rehosting solution is the right thing for everyone. In our case, the results were spectacular but that by no means guarantees success for anyone else attempting a similar project. Lastly, I have no desire to get into a religious war of mainframe vs. unix. I am platform agnostic. We found something that got us to where we wanted to get to and saved us some decent amount of $$$ along the way. I just presented our case study at a Gartner application development summit and the responses were tremendous. If nothing else, I think our project demonstrates that it is indeed possible to migrate multi-thousand MIPS mainframe environments to a UNIX based system with similar levels of reliability and performance. Sangho Yoon </snip> ---------------------------------------------------------------------- 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