Sent from Yahoo Mail for iPhone

Begin forwarded message:

On Friday, October 22, 2021, 9:39 AM, Bill Johnson <mellonb...@yahoo.com> wrote:

Fintechs will never replace or even challenge banks.
Why the future belongs to banks, and not Fintechs 
  
|  
|  
|  
|   |    |

  |

  |
|  
|   |  
Why the future belongs to banks, and not Fintechs
 
With emerging Fintech services taking the world of finance by storm, it might 
seem like banks are in...
  |   |

  |

  |

  


Sent from Yahoo Mail for iPhone


On Friday, October 22, 2021, 6:35 AM, David Crayford <dcrayf...@gmail.com> 
wrote:

On 22/10/2021 5:59 pm, Radoslaw Skorupka wrote:
> W dniu 22.10.2021 o 03:12, David Crayford pisze:
>> On 21/10/2021 7:31 am, Bill Johnson wrote:
>>> I do almost everything important via app with a mainframe on the 
>>> back end. Banking, health records, retail shopping, insurance 
>>> claims, investing. And with very high confidence the transactions 
>>> are secure, fast, & always available. Not sure how more modern the 
>>> mainframe could be actually.
>>> Bill J
>>
>> I would suggest that almost all of those companies (banks, retail, 
>> insurance etc) that run mainframes run a significant portion of their 
>> IT systems on distributed systems. I recently met a software 
>> architect from a bank who told me that the explosion of internet 
>> banking app usage has pushed up the TCO of their mainframe to 
>> unacceptable levels. 70% of those transactions were reads (folks 
>> checking their bank balances on their phones). Their solution was use 
>> Apache Kafka to replicate the data to Apache Cassandra and only hit 
>> the mainframe for writes.
>
> BTDT, but not on Kafka. Indeed, the "cache system" ran up to 40% 
> transaction.
> Sounds good? Imagine: 40% less workload...
> NO
> There are light and heavy CICS transactions and cached ones were light.
> Proof: some day the cache stopped working due to system failure. 
> During rush hours.
> Observations from mainframe side: number of transactions per second 
> blew up. Average response time ...decreased a little bit.
> I was really happy reporting it to management.


The link I provided was Australia's largest (and a world top 20) bank 
with millions of transactions a day. They're not stupid, production 
technology choices are critical which is probably why IBM have spent $$ 
making sure Kafka runs ok on z/OS.

Caching isn't a new idea. It's a common CICS design pattern using TS so 
you don't have to make an expensive call to DB2 or IMS. The customer 
solution is not call the mainframe for read transactions. It's not 
uncommon, it starting to become pervasive. Writes are a different matter.


>
> However the management was not happy because of that, just because 
> they want to switch the mainframe off.  Nevermind, the new transaction 
> system has response times 35-140ms (compared to 4-5ms on mainframe).
>
>
>

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN







----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN

Reply via email to