Friday, January 4, 2013

What is SAP HANA ?


In this article we will look at SAP HANA which is the most talked about product of SAP after its realased to customers on Dec 2010.
SAP’s High-Performance Analytic Appliance(HANA) exploits the advantages of In-Memory Computing that allows the processing of massive quantities of real time data in the main memory of the server to provide immediate results from analyses and transactions by combining SAP software components optimized on hardware provided and delivered by SAP leading hardware partners.
Reminds me of movie “3 Idiots” . Perfect definition to score 10/10 in any exam or interview. :)
Lets make it a little simple.
SAP HANA is a combination of hardware and software specifically made to process massive real time data using In-Memory computing
We managed to get a sneak peak of HANA hardware in Teched 10 at Bangalore.
SAP-HANA

Need of SAP HANA:

Quoting  Vishal Sikka “ If a business asks a question and gets the answer after 3 days probably one would even forget what was the question was after 3 days. “
So one of the key challenges IT is facing now is not only the capability to extract reports on huge amount of real time data which keeps growing in a exponential rate and from different sources but also analyzing it in different perspectives, that too in seconds.
Most of the Official SAP presentation refers to a popular term in Finance “ TED Spread “ , for explaining it , which is a measure of credit risk for inter-banking lending.
12179_b
Aftermath of the Lehman bankruptcy, TED Spread hit a record spread of 465 basis points. That magnitude of spread indicated a virtual shut down in interbank lending. All financial organizations use the TED spread as an indication of easing or tightening credit markets.
It might be a little difficult to digest it in the first go and since I am not a Finance guy so probably you can Google around to find more about TED Spread.
Summarizing the whole thing in a single sentence would be that every organization needs to make better decisions faster not only on real time data but Real real time data which keeps growing in a exponential rate and from different sources almost instantly.
When I say real real Time Data for a retail shop perspective the POS (Point of sale) data would be available for analytics even before the customer leaves the retail store.

Motivation for Architecture of SAP HANA

Most of the chip maker companies have been focusing on building multiple core as there is limitation on increasing the Clock Speed as  the more is the Clock Speed the more heat it emits while processing which results in expensive ways to control the temperature. If you observe the last decade the Clock Speed has not made any impressive change.
2002                                                         2006                                                         2010

Chip
Cores/CPU -   1 Core                                      4 Cores                                                                 8 Cores
Clock Speed – 1.8 GHz                                   1.6- 3GHz                                                             2.26 GHz
Software has always been made for the Hardware which these Top chip makers design.
There is always a need to build Software which can fully exploit the capabilities of the Hardware design.
So let’s have a look on both hardware and software features one by one.

HARDWARE DESIGN:

Huge amount of data is divided into multiple sets which are then crunched separately by the Blades as shown below.
blades1
blades2
Pic : Data is divided into 4 blades with 2 standby blades
The Blades are composed of multiple CPU’s per blade and each CPU has multiple cores per CPU.
This means if you say for example 8 cores per CPU and 4 such CPU’s per blade. So mere 4 Blades will have 128 cores crunching data in parallel.

SOFTWARE DESIGN:

HANA stores data in Column wise for fast computing. The below diagram compares, how data is stored row wise and column wise.
software-hana
For example: If system wants to find aggregate of the second column i.e. 10+35+2+40+12.
In Row wise: The system has to jump memory address to collect subsequent values for aggregation.
That is data records are available as complete tuples in one read which makes accessing of few attributes expensive operation.
In Column wise: A single scan would fetch the results much faster.

HOW IT WORKS:

saphana
SAP HANA can be used without disturbing our current IT landscape. In the above diagram show that data in the database can be replicated near real time into HANA and can be used for reporting with a number of BI tools directly sitting on the top of the HANA.
Post a Comment