I recently worked on implementing SAP HANA at a major retail and investment banking customer. This implementation consisted of a 2TB and a 4TB deployment of HANA and hosted only the SAP Business Warehouse environments that were growing at an average of 450GB (IBM DB2 space) per month. Growth was influenced by the amount of data being migrated from old Legacy systems and some organic growth from new business.
DATA EXPLOSION
The customer was experiencing EXPLOSIVE data growth within their Business Warehouse environment as they were engaging in an infrastructure modernization and transformation process that spanned multiple years (past and future) which included moving a lot of data off old legacy mainframe systems for multiple business environments.
Additionally, a large number of new accounts were being originated via campaigns to win new customers and with the support of their SAP Sybase Mobility Banking deployment they were able to reach both urban and rural areas for new business. As a result an explosive amount of new transactional data fed through to their BW environments from various source systems.
COMPLEX ENVIRONMENTS & COSTLY INFRASTRUCTURE
As with most banking customers, system environments are rather complex and interwoven with custom and 3rd party solutions. We also had various SAP solutions interlinked with each other for example Banking Services which include Branch Accounting, Bank Analyzer, Account Origination, Loans & Home Loans and CRM, ERP, BW to name a few, plus downstream systems dependant on BW.
The need for SAP HANA was highlighted by operational challenges due to batch windows (overnight and weekend) becoming uncomfortably tight. The bank’s system and DBA teams have exhausted performance and tuning options. All processes have been extra finely tuned so much that they were finally succumbing to disk speed as the ultimate bottleneck.
This and the fact that they had such a massive and costly storage footprint on mainframe technology with an alarming growth forecast were just some of the reasons why they looked at alternative technologies to help ease these pains and to assist in their effort to modernize their core banking systems.
GOAL DRIVEN
Having worked with this customer on various projects in the past gave me an understanding of a variety of goals that were important to achieve including:
- Their pursuit for REAL-TIME visibility of data across business silos which in itself drove the very transformation of technology they invested in and continue to invest in
- To get the ultimate performance out of their systems which would enable them to RUN BETTER, save cost and be more PROFITABLE because better decisions can be made FASTER, when it matters!
- They continue to compete for the accolade of being the best and biggest and are still striving to deliver a better service to their customers, right NOW! By improving the customer experience, they win new business.
By the time I joined the project, the customer already opted for HANA and we were excited to be involved and part of the team to plan and execute the implementation!
The solution consisted of HANA Appliances which hosted multiple environments. An upgrade of the Business Warehouse to 7.3.1 suite was needed and SAP Sybase IQ NLS as the Nearline Storage Solution was going to be implemented as the complimentary technology to HANA for keeping COLD data, readable and still online but as a cost effective way to help reduce the memory footprint in HANA.
IMPACT on the SYSTEM LANDSCAPE
The implementation of BW on HANA with IQ NLS had to be planned very carefully in order to reduce the risk and impact on the entire system landscape both for BW environments, upstream and downstream systems and other business units developing projects at the same time!
To mitigate the risk, we identified the processes required to get it implemented and unpacked the approach into various work streams in order to fully understand and plan the steps and sequence of execution. We had to take into account (in no particular order):
- Feeds from SAP Systems, like Banking Services, CRM and other Source Systems into BW and the impact to them
- Impact to batch and downstream systems depending on the batch
- Classification of BW data into HOT, WARM and COLD data as this impacts sizing of both HANA and IQ
- Upgrade BW to at least version 7.3.1
- Multiple DEV and TEST environments, UAT, Pre-Prod, DR and which will be done first and which will be retrofitted after going into production.
- Introducing X86 infrastructure to an environment that is mostly AIX and Sun Solaris
- Availability of Linux skills
- Identifying Infrastructure and Storage Requirements
- Design & Procure, Install & Configure all in time for each deployment per environment during the program lifecycle
- Business Warehouse application servers on AIX and re-platforming them to LINUX
- Prod-like Environments needed for rehearsing the sequence of events and performance tests
- Migration of HOT and WARM data into HANA from IBM DB2 databases
- Migration of COLD data via NLS from IBM DB2 databases into IQ from the current BW IBM DB2 environments
- Test Cycles per sequence that impacts the program timeline
…to name but a few
The particular business processes included: (but were not limited to)
- Retail / Core Banking activities including Accounts, Payments, Loans, Home Loans, Credit Cards
- Business to Business banking
- Mobile, Cellphone and Online Banking
- The Corporate Investment Banking arm
SAP RESOURCES
The project team included multiple SME’s (subject matter experts) a SAP HANA SME, a MIGRATION expert, SAP Sybase IQ (myself) and SAP Netweaver BW SMEs.
My main area of expertise is SAP Sybase databases like Sybase IQ / ASE & SAP HANA.
For this project I was specifically tasked to lead the SAP SYBASE IQ NLS solution stream, responsible for the architecture, infrastructure build requirements through to implementation and configuration and migration of data into IQ via the NLS add-on in BW.
I was also part of a Batch Re-design team who were tasked to look at the current processing to see how much of it can be changed into real-time processing.
We all worked together to unpack all areas of the BW on HANA & IQ NLS deployment including Project Scope, Implementation and Timelines as well as discussing technical product details in various workshops with input from various technical teams in the bank.
HOW DID WE HELP
Our team shared our expert knowledge during the various workshops, so that our SAP solutions could be understood better and good decisions could be made around timelines and program execution.
This implementation aimed to bring an enormous benefit to the customer in the following areas:
- A stunning performance improvement as the mostly accessed or HOT data is in memory, compared to where it was residing in DB2 row based database technologies on traditional disk based storage before. In HANA WARM data is on SOLID STATE disk in the HANA appliance, COLD data readable on-line in IQ.
- A Significant reduction in storage space and costs
- An immediate improvement on the batch window as a combination of speed and compression would make a dramatic reduction on the time that the batch takes to run, having a freed up window available to run extra jobs that would help them transform and migrate away from legacy systems quicker.
- If a decent compression ratio is obtained, a consequential benefit in performance for queries, since fewer data and index pages will be read.
- An SAP IQ NLS solution that aims to keep COLD data still online while keeping the HOT data footprint in HANA at a manageable size.
REWARD
The most rewarding aspect of this project for me would be to see the customer achieving the ultimate in performance whilst saving huge amounts of space and storage cost and to witness how HANA enjoys further uptake in other areas of the business which will enable us to do this again for others. At this stage some of the other teams in other divisions of the bank already want a piece of this HANA appliance!
See attached photos as an example of the appliance
My colleagues at another customer obtained the following statistics with BW on HANA and IQ as near-line storage solution. I include some slides below as an example of the performance gained there, for example an Adhoc Query accross 78 Billion Rows of data coming back in 13 seconds on HANA!
Please contact chris.bezuidenhout@sap.com for more information on that particular POC.
Thank you for reading my blog :oops:
LIEZL NEL
Banking Services
SAP South Africa
