When CFO Innovation spoke to German enterprise software giant SAP in 2012, it was getting ready to enable hosting of ERP in the in-memory machine HANA.
“HANA can store 100 terabytes, which is just a mindboggling amount of information,” said Andy Hirst, Vice President for Banking Solutions – in addition to improving query performance 100,000 times when the data is stored in-memory, not on hard disks.
Four years on, the ERP system S/4HANA has been installed in HANA machines by more than 3,700 customers worldwide, says Rohit Nagarajan, Vice President, Database and HANA Cloud Platform, at SAP. And “HANA can be your data platform or your advanced analytics platform,” he told CFO Innovation’s Cesar Bacani. Edited excerpts of the interview:
Because of the digital economy, organizations will have requirements around real-time processes, consumer-based user experience, and so on. That wasn’t being developed in the traditional database platform, so we came up with HANA from the ground up
Tell us about HANA.
Most databases are created to write or to read data. HANA can do both – and do it equally fast. It stores all of the data that you have in the organization in your DRAM, your random access memory, which means it is extremely fast.
And it’s not just a database. It’s a platform, because what it also has are a lot of engines like graphing engines and predictive engines, which act on the data in-memory.
That gives a lot of benefits to an organization around simplifying the IT landscape because all of the organization’s data can now be stored in-memory. You have one copy of the data, and you can perform all kinds of complex advance analytics on that data.
The reason we came up with HANA was very simple. We knew that the digital economy would require a next-generation data platform. Because of the digital economy, organizations will have requirements around real-time processes, consumer-based user experience, and so on.
That wasn’t being developed in the traditional database platform, so we came up with HANA from the ground up. HANA is now the world’s leading in-memory platform. It has over 10,000 customers.
Is HANA different from S/4HANA, or are these terms interchangeable?
S/4HANA is a next-generation application written specifically for HANA. S/4HANA runs only on HANA.
An analogy I’d give is: We first designed an extremely powerful engine, a new engine completely out of the ordinary, which could do things that no engine before it could do. Then we realized the power of this engine through a car designed specifically for that engine, which maximizes the capability of that engine.
We’ve rewritten our ERP suite to leverage the capability of HANA. The fact that all the data is in-memory meant we could simplify the data structures. We could give users access to extremely granular pieces of information in real time.
But I can run other applications on HANA, not just S/4HANA?
Yes. We have a lot of customers who use it for very innovative applications which are not S/4 related. So HANA can be your data platform or your advanced analytics platform for the custom application.
Say, for example, you’re a retailer or a telecommunications operator, and you want to do churn analysis on which subscribers might leave. You could write your own custom application, which leverages the power of the predictive engine within HANA, to write your own capability and deliver your own results.
SAP is a software company. Does it now manufacture hardware in the form of the HANA machine?
No. We have certified partners from HP to IBM to Dell to Cisco, Huawei and Lenovo – our customers procure certified HANA appliances [from them]. We also have the HANA Enterprise Cloud, which is the SAP HANA certified cloud service.
Many customers choose to deploy this HANA service. We also allow our customers to run HANA on AWS [Amazon Web Services]. Customers have a variety of deployment options.
Suppose I already bought on-premise Oracle ERP which runs on Oracle databases and servers. I then buy a HANA appliance because I like the speed made possible by the in-memory computing. Can I load the Oracle ERP into this HANA machine?
From an ERP perspective, Oracle doesn’t certify their ERP on HANA . . . The Oracle ERP basically runs on Oracle databases. It doesn’t run on the HANA database. If I’m an Oracle ERP customer, I can’t run my Oracle ERP on HANA.
If I already have an ERP that runs on a server, not on HANA, what I can do is buy a HANA machine and then put it on top of that. What HANA will do will be to extract the information, put it inside its in-memory, and then do it much faster
Let’s take a company which currently doesn’t have HANA. They might have an Oracle database, they might have Cognos or Hyperion. They have an ERP system where they have transactional data.
Then they’ll take the data and move it to a business warehouse, where they’ll perform some financial reporting on it, monthly and quarterly reports. Then they take the data to another data warehouse and then perform some predictive analytics on it for future-looking advanced analytics.
That basically means you’re moving data from one place to another. You’re losing the granularity, the richness, of the data as you move, there are costs and overhead around the integration. That’s what we were looking to avoid.
In a HANA environment, all of the data resides in HANA; you can do your ERP, you can do your analytics and your predictive analytics. So the concept of having an Oracle alongside it, or a Cognos alongside it, that almost becomes irrelevant.
You’re talking more about a massively simplified self-containing platform, where all of your data goes in and you basically perform all the analytics you need on it.
If you want to access the data through Business Objects, for example, a lot of our customers do that. We have customers using BI and Lumira and even Tableau to visualize the data. But all of the actual processing is done [in HANA]; you do not need any further data platforms on top of HANA.
HANA can abstract all of your other data sources. But effectively our customers use HANA as the single simplification data platform for their organization.
So I will have two systems, two sets of servers. One would be the normal data servers and then the other would be the HANA.
If I already have an ERP that runs on a server, not on HANA, what I can do is buy a HANA machine and then put it on top of that. It will run parallel to that; it will continue to do the ERP things that it does.
What HANA will do will be to extract the information, put it inside its in-memory, and then do it much faster than it would have been if you were using outside of it.
What we see customers doing, in the short-term, they have two sets of servers. But as they use HANA increasingly for their business process, all of the new data goes into HANA, and less and less data seem to be going into the existing Microsoft or Oracle servers.
Picture a box called HANA, which sits on top of every other investment that you have made in terms of databases, effectively abstracting [data]. If you need to ask a question, you ask HANA.
HANA will go in there and find out where it is currently stored in terms of the database underlying it, and give you the answer. New data goes into HANA, which makes all the new data accessible in real-time.
When it comes time to perform your technology refresh on the server Oracle or Microsoft, a lot of customers have a decision point where they say: Why don’t we just increase the HANA server as opposed to Oracle, since all of the data is going to be there anyway and all of my business insight is coming from that layer?
They often turn off the servers underlying that, which is resulting in a significant lowering of their total cost of ownership [TCO] because they no longer have multiple servers. They’re still getting the business benefit, but they are decreasing the total cost of ownership.
We have done TCO studies that clearly indicate a HANA-based platform has significantly lower [total cost of ownership] compared to a non-HANA-based platform because we avoid redundancy of data. You don’t need additional servers for your planning and your advance analytics; it’s all in one server.
We have very standardized pricing, especially for S/4HANA, which is something that we share with our customers on the basis of their landscape. But we don’t share that openly, in terms of the price list
What happens to my Oracle ERP software if I turn off the underlying server? I will need to deploy S/4 or another HANA-compliant ERP?
This is a path that [our customers take] to go to S/4HANA, eventually.
They start with HANA, they [put it on top of the existing ERP], and then when the right point comes – that could be a technology refresh, it could be another event in the company around M&A, which necessitate the need for a new ERP – they use that as a triggering point to move to S/4HANA.
Let me ask you about cost. I’m assuming that HANA, the platform, versus S/4HANA the ERP system, there’s a difference in cost?
We have multiple licensing methods. The HANA platform can obviously be procured separately, which means it has separate pricing from the S/4 HANA pricing.
Do you have a range of the pricing, just for CFOs to know how much investment they should set aside if they are interested in HANA and S/4HANA?
We have very standardized pricing, especially for S/4HANA, which is something that we share with our customers on the basis of their landscape. But we don’t share that openly, in terms of the price list.
It’s a very standard pricing, which is very competitive with the other vendors in the market and it’s offered as separate for S/4 and separate for HANA.
We meet with CFOs and present the business case for HANA all the time. There are two elements for S/4HANA: the topline benefits, which are the process improvement, the faster processes, the CFO dashboard, the real-time insight, the fancy user experience; and the second is the reduction in total cost of ownership.
The lowering of the total cost of ownership comes from the reduced memory footprint, because HANA stores everything in-memory in a compressed fashion, and the overall reduction in the growth rate of servers required, because the data can be stored in a smaller footprint.
Into that calculation, we also input our license considerations, which [our studies have] always shown to be a lower TCO than any comparative data platform.
Do I even need to buy my own HANA machine? Don’t public cloud companies and SAP itself allow me to rent on subscription basis?
Customers have a variety of options for their HANA deployment. Some of them choose to buy their own servers and do it on-premise. Many others use the SAP HANA Enterprise Cloud, which is our cloud platform for customers who want us to host their HANA for them, instead of them hosting it.
We also have partners who do that, large and small. And yes, Amazon and [Microsoft] Azure, they both have HANA servers that customers can choose to lease. It’s an indication that the broader partner community, Amazon and Microsoft included, realize the scale at which HANA is growing.
Customers make a choice based on many issues, the data residency issues…each customer has his own thinking behind whether they want to keep it on premise or cloud, for various reasons. But they all have a choice because we provide that to them.
Can you give me a rough idea of the take up for S/4HANA?
The number of S/4HANA customers is 3,700 as of last count. What’s interesting about that is, in Q2 alone, we added over 500 customers. That’s showing the significant growth rate at which customers are coming on board.
For HANA [the platform], I think we stopped keeping count after December. It was over 10,000 customers in December 2015.
Some of them might choose on-premise servers. Some of them might choose to go with the HANA Enterprise Cloud. Some might choose to go with Amazon and Azure.
We work with very, very large names across Asia Pacific and Japan. They use HANA to power both their core ERP business processes as well as a lot of innovation use cases . . . It’s just that we cannot publicly reference them
And how many customers are in Asia?
Anecdotally, we’ve got extremely strong uptick in the Asia Pacific Japan market, but we don’t, as corporate policy, break down these numbers [by geography]. Two of our large reference customers in this region are Asian Paints in India and La Trobe University in Australia.
Asian Paints is a large manufacturer in India that has come on this journey with us from HANA through to S/4HANA. They went from project start to light launch in something like seven and a half weeks. And they have seen 45 times faster processes on sales and marketing, 50% faster backing processes and 60% reduction in database requirements.
All of those technical benefits are powering a lot of business benefits they have around better customer experience because of real-time interaction. They’re able to do real-time campaigns and offers to their customers.
La Trobe University is also having significant benefits. They have seen 20% increase in efficiencies, seven times improvement in business processes, and business-specific benefits around allowing the admin staff to forget the overhead and focus on the teaching and the core tasks.
And they have [acquired] some very innovative ways of tracking student performance and student funding in very challenging higher education environment.
The Toyotas and the Samsungs and the HSBCs, are they not on the HANA platform?
Because of the fact that a lot of the capability that customers gets delivered because of HANA provides them with significant competitive edge, not all of them are publicly [referenced].
Suffice it to say we work with very, very large names across Asia Pacific and Japan. They use HANA to power both their core ERP business processes as well as a lot of innovation use cases.
Like I said, 3,700 customers, just 500 of them in the second quarter of the year, show the rate of uptake. Obviously that means a significant portion of our installed base is choosing to take that journey, and those would be the larger organizations. It’s just that we cannot publicly reference them.