In-Memory of an Eight Hour End-of-Day – Transforming Banks in Realtime

In an effort to speak to other innovations, aside from the popular mobile banking, I thought I would talk about in-memory functionality.

There was a recent article in Forbes about SAP’s HANA product and the game changing power of speed in business. I was at SAP’s SAPPHIRE conference in May of 2010 when they announced the in-memory initiative. About six months before that conference I was at the SAP headquarters in Waldorf Germany and heard rumours of this “game changing” technology. Now I am actually starting to get questions about how the current package of SAP-HANA could help banks in their transformations.

I’m going to talk about the SAP’s HANA offering, however the concepts behind in-memory computing are more important than a brand-name and the overall benefits to a bank is what should be taken away from this blog.

In-Memory: Basically for SAP’s in-memory solution to work you need three things:

1. HANA Database  – This is the new brand name of the SAP in memory data base (IMDB)

2. HANA Appliance  – This is  new hardware.  SAP and their hardware partners have gotten together and basically created a new type of server that replaces the old style iron hard disks with new electronic storage. This is in essence the real speed behind the solution. Storing and retrieving data in the electronic format without the spinning disks saves an incredible amount of time and effort. Once the HANA Database is installed on a certified HANA server it becomes a HANA Appliance.

3. HANA Studio – The toolset used to administer and model the HANA Database is called the HANA Studio

Situation Normal

So you are a bank and you are currently running your daily operations. At the end-of-day you close your books and start to run what can be an extremely long end-of-day process to determine the overall position of the bank, update the general ledger, and produce reports used the following day to manage the bank. In some cases, OK in probably all cases, you will dump a large portion of your data from a number of disparate systems into an enterprise data wearhouse where it will be further  transformed to support other processes and analysis. In this standard method of operation you wait for your data and then, based on the route it takes to get to you, you may question the validity of it.

With In-Memory

Real time transactional data is stored in lookup tables in the HANA Database on the HANA Appliance. There is always going to be some latency between your operational systems and the HANA Appliance so it should be considered near real time. The beauty of this is that you can load SAP and Non-SAP data. Once transactional data is stored in the HANA Database, which was previously modelled using the HANA Studio, you can currently use SAP Business Objects or Excel to produce real time reports without taxing your operating systems. This is where the dramatic change in how a bank does business can occur. Some of the ideas I have around this use include:

1. Realtime FOREX position across an entire bank

2. Treasury and cash management in realtime

3. Realtime operational risk monitoring and reporting

4. Realtime portfolio stress testing

5. The creation of realtime cash management services to be resold to commercial and corporate clients

6. AML monitoring in realtime

7. The ability to pull any current end-of-day report at anytime during the day

Granted these things do not come out of the box and require a not insignificant amount of effort to stand up, however in-memory computing is a relatively new tool that can be leveraged to dramatically transform how a bank does business.





16 Responses to In-Memory of an Eight Hour End-of-Day – Transforming Banks in Realtime

  1. The point is not that in-memory speeds up the end of day, although when you move previous end of day processes into realtime with in-memory you are by default shortening your end of day by removing processing time. The point is that in-memory makes the end of day irrelevant for the processes and analysis contained within it in that you no longer need to wait for the end of day to complete to get a clear picture of yesterdays data. You already got yesterdays data, you got it in near realtime yesterday, and you used it immediately you no longer care about end-of-day completing.

  2. Can you critique the differences between real time processing and memo posting? Also, given the time shrinkage between transactions and real time balances, seems to be a field day for wrong doers..

  3. In a practical application of in-memory the financial institution would have to determine when a transaction would be included in their in-memory solution. From my point of view you would want to take the memo posting as it occurs and divorce your in-memory data from your end-of-day. As I have tried to say in-memory is not currently a replacement for your operational data bases and would not negate the need for an end of day process, down the road it could. What it allows you to do is to continuously dump fresh transactions into a near realtime data store that can complete analysis of the data as the data is received giving the financial institution a clearer line of site on their operations, risk, and exposure as the day goes on not having to wait for the end-of-day to complete to get yesterdays news.

  4. hi George: could you provide mnore detail information about that? sound very interested but I apreciate if you can shared some more detail information.
    another important point in the end of day process is the inetgration with other systems or other entities, when the bank must wait for some result, to continuethe crtical way.
    thank you.

  5. Currently I am not aware of any bank with an in-memory solution in production. Alberta Treasury Branch, a regional bank in Canada, that went wall to wall SAP recently has opened an banking innovation centre and I know the SAP Hana will be one of the first implementations and targeted for production soon after. Having said that it is only a matter of months before we start seeing Hana in other banks.

  6. The in-memory concept for core banking solutions was first developed and successfully deployed by Financial Network Services back in the mid to late 80’s. FNS Bancs, now known as TCS [email protected] following the acquisition of FNS by TCS in 2005, is used by banks around the world including one of the biggest by infrastructure the State Bank of India.
    FNS Bancs provided on-line real-time positions by entity (branch, institution) and near real-time positions by currency, product and so on. In the late 90’s the 24/7 version of FNS [email protected] was launched removing the dependency of an overnight batch and allowing the system to be on-line 24 hours a day and thus support the needs of the proliferation of banking channels to have access to a centralised and thus single view of arrangements, e.g. account balances, facility utilisation etc. FNS [email protected] supported a logical end of day which when it was deployed for Banks running across multiple time zones was always an interesting experience. This was a rare example of where a very technology dependent solution was very closely aligned to a business need.
    Its good to see that what was at the time considered to be a true innovation has been improved upon by SAP. It is a stretch to describe SAP HANA as innovation as it is arguably simply using advanced technology to improve upon an existing idea.
    As for transforming Banks, my question is “transform them into what?”, in my experience gained from providing banks around the globe with real-time on-line access to information they have still remained Banks albeit they became more profitable as their improved information systems helped them to grow revenue and deliver productivity improvements and associated cost benefits.

  7. Hi,
    Finacle core banking solution supports 24 x 7 operations, without having any stand by time for end of day operations. As such all operations are exceuted in real time, now major core banking vendor supports this in memory posting.

  8. I always spent my half an hour to read this weblog’s posts every day along with a mug of coffee.

  9. In the brochure of finacle it says “24 * 7 capabilities to operate round-the-clock through branches and channels even during Job execution and day-end operations”
    That’s exactly the same as SAP can offer since many years even without the HANA Technology.

  10. You are correct. The problem is even thought they can continue transactional postings they are still running end-of-day processes that are taking 8 hours or more to complete. The ability to continue 24 by 7 posting has nothing at all to do with closing the end of day. The benefits of shortening time it takes to generate business intelligence and action the same from current day transactional processing is the key. Proactive action based on real-time or near real-time data is something that Finacle cannot address. SAP Hana with the correct source data from real-time systems can provide views into the business as it happens real time!

  11. Hi All, I have 7 million accounts stimulated on my local oracle dev db .
    I had designed EOD algoritham I am completing some operations for all 7 miilions in 0.3 seconds –
    Here are the operations I am completing –
    1) NPA Marking Unmarking
    2) Simple Interest Calculations
    3) Penal Interest Calculations
    4) Account Dormancy and marking again back to Active.
    My Hardware is –
    Testing Hardware Normal Laptop with Windows 7 as OS and Oracle 11g as a Database.

    Processor: Intel (r) Core (TM) i5-4200U CPU @ 1.60 GHz 2.3 gHz
    Installed Memory(RAM): 4.00 GB (2.98 GB usable)
    System Type: 32-bit Operating System

    I can put up this solution to any CBS . Let me know .
    This is Manoj J Patil
    [email protected]

  12. Also I can Build 24 Hrs processing capability and very short EOD in any Core Banking System
    Right Now we are also executing Standing Instructions Stress Testing for all accounts – 7 million and the results are unbelivable !!! Let me complete some homework and comeback on the timing for trn posting on normal hardware as well

  13. Just to Update we can extend our fast processing to various other systems – payment processings, Reports involved in the same and systems like ALM, AML and various steps for the same.
    Looking out for Implementation of the same

Leave a Reply

Skip to toolbar