SAP HANA and Hadoop – a great partnership
SAP HANA can benefit the financial sector in areas such as sales, marketing, risk management and financial accounting. However, besides the impressive possibilities that SAP HANA presents, we also have to take into account the evolutionary process it has undergone to arrive at where it is now, looking at both the strengths and weaknesses that it currently presents.
In the last few months, SAP HANA has seen a great leap forward in its development tools. It is now much simpler to use. Today we have a central platform that remains as solid and stable as before, while the development platform has been significantly improved. Its basic functionality can be executed in a shorter time – something which enables more efficient development. Nevertheless, one still has to remember that bug fixing and fine tuning the application is still a challenging task and should be planned for.
Can market leaders in Big Data and in-memory technologies collaborate to bring about high-performance solutions? At GFT we are certain they can. Based on our experience, we can be sure to achieve our objective of creating an integral solution for managing large data sets, by combining the strengths of SAP HANA and Hadoop.
SAP HANA and Hadoop come together
SAP HANA is particularly efficient at making real-time decisions and provides support systems for decision-making. It is also very good at managing large amounts of data, although not yet at the same level as Hadoop.
On the other hand, Hadoop enables large amounts of data to be stored arbitrarily in an efficient way. One of its primary strong points is that it allows us to find a needle in a (huge and unstructured) haystack. In short, it carries out real and complex data mining, always running in batch-processing mode.
The table below shows a comparison between HANA and Hadoop based on the following four criteria:
• Time Scales and Interfaces:
- SAP HANA is in its element when it comes to real-time environments, though is also able to operate in batch-processing mode. It allows access to stored data through the use of standard interfaces, as well as offering customised development.
- Hadoop runs only in batch-processing mode and requires additional effort in order to make the stored data available to other components of the IT infrastructure, usually through the use of specific customised interfaces.
• Data Volume and Structure:
- Hadoop allows you to manage huge volumes of data. It is most efficient in handling unstructured or semi-structured data.
- SAP HANA allows you to manage large volumes of data and is most efficient in handling more structured types of data.
• Total Cost of Ownership:
- SAP HANA is a commercial solution that requires a purchased licence. It comes with a high set-up cost.
- Hadoop is non-commercial, Open Source software. As with all Open Source products, it has a very low set-up cost. However, in order to build a production infrastructure it is necessary to make additional investments to ensure the high quality of the service required.
• Deployment:
- SAP HANA requires a specific appliance and allows limited use of the clusters in parallel processing, usually in a local area.
- Hadoop allows an extremely high level of parallel processing in clusters, which can be located remotely in the cloud.
After testing both technologies in a pilot development, GFT strongly supports the combination of SAP HANA and HADOOP as complementary technologies. With this joint venture we can face the challenge of managing large volumes of data, both in terms of speed and on a variety of scales. The following chart compares volume and velocity against volume and variety for SAP HANA and Hadoop, as well as the combination of both. Evidently, a combination of the two enables the total spectrum of possibilities to be covered, and therefore produces the optimum results.
In conclusion, combining both technologies can leverage their individual strengths allowing them to build a comprehensive Big Data solution. SAP is currently working on integrating SAP HANA and HADOOP.
