7 Years of HANA: A Perspective on the Past and What’s Coming Next
I remember the first SAP HANA keynote Hasso gave in SAPPHIRE 2010 in Orlando, back when we still referred to it as NewDB.
In it, Hasso describes a six step process that allows SAP customers to move their systems onto the new database without disruption. Sorry about the fuzzy picture, it is an old video!
Three things struck me right away:
First, the SAP HANA services market was going to be big business over the next ten years. Every customer was going to need assistance upgrading their systems and streamlining business processes. In my role as Business Analytics & Technolgy practice lead at SAP Services partner Bluefin, that sounded like a big deal to me.
Second, collapsing the transactional and data warehouse solutions into a single system of record and analytics might be harder than it sounded. A good EDW has complex business rules and data from multiple systems.
Third, whilst building a sidecar database to accelerate reporting was conceivable, building a complete RDBMS complete with ANSI SQL, ACID compline, high availability, security and enterprise reliability sounded like a tall order. There hadn’t been a new kid on the Enterprise RDBMS block since Microsoft in 1989, and it wasn’t until 25 years later that serious enterprise systems ran on MSSQL. Building an enterprise-grade RDBMS is serious business.
I was wrong on the third point: SAP HANA 1.0 was released in November of 2010, and some 7 years later, it is capable of running pretty much any SAP database in the world. As SVP of SAP Solutions at Bluefin, a global consultancy and SAP partner, I’ve personally seen the largest and most complex systems in the world run more simply, with reduced cost of ownership and all the benefits of SAP HANA.
In fact, I haven’t met a SAP customer in the last year who hasn’t got SAP S/4HANA – SAP’s 4th generation ERP – on their mind. Many have already made the jump, and most others have a plan to get there. For some customers with complex processes and large amounts of custom code, this requires careful planning, but these projects are underway.
And so it happened on a warm summer’s day that I talked to the SAP leadership team about their vision for HANA’s next chapter – and was excited enough about what I learned to join SAP’s Database and Data Management business.
In short, we have fantastic ambitions as to the future of SAP HANA. In particular I think there are three things that excite me right now.
The trends in the public and private clouds have provided deployment options we couldn’t have dreamed of five years ago. Amazon just announced 4TB single systems, Microsoft provide bare metal systems that are even bigger, and Google just got out the starting blocks in May, and are one to watch.
Private cloud vendors like IBM, SAP HEC and Virtustream offer more customized deployment options with enterprise security which trounces what most organizations can afford in their own datacenters. Competition and choices make this a great market for buyers.
In addition, we see customers moving towards hybrid deployments, for example deploying cloud solutions on-premises, which is one of the fastest growing market segments.
Second, at SAP we are committed bring HANA well beyond being a database for SAP applications like S/4HANA.
If you think about where the HANA investments went over the last five years: building a stable database that can run almost any SAP application, on both Intel E7 and IBM Power, from most of the major and cloud vendors, on both of the major Enterprise Linux distributions, various hypervisors like VMWare, Xen etc., with sophisticated development and administration tools, high availability and disaster recovery, integration with the major administration and backup tools. We have native OLAP, Graph, Text, Search Spatial, Streaming and Predictive. This is an enormous amount to achieve over 5 years.
In doing so, we built the only data platform which could run S/4HANA, and that’s the great thing: because it runs SAP applications, which are amongst the most demanding business applications in the world, it can also run just about anything else.
With SAP HANA 2.0, this energy has been redirected into expanding the use cases: Search, Text, Graph, Predictive. We allow you to double the compute power by using Active-Active. The platform has been opened up with Bring Your Own Language (BYOL) on the XSA Application Server. We have built data management into the platform with data tiering, integration and quality capabilities. In short – we are building a platform which any developer, entrepeneur or business unit will delight from.
Performance, Quality & Security
Third, the focus has to continue to be on performance quality and security. Several database vendors struggled in the past to be considered enteprise grade, and in 2017, customers have very little patience. If you read the new feature logs for SAP HANA 2.0 you will see this in action.
There are improvements to HA, DR, Workload Management, Administration, Monitoring, and the development team measures stability as a priority. Undo’s Live Recorder is used for test automation.
We have also seen significant performance improvements, even compared to earlier versions of HANA. In particular, whilst earlier versions of HANA were outstanding at analytical workloads, with HANA 2.0, we also have class leading transactional performance, even with operations on a single row, which is a key strength of a traditional RDBMS.
I join SAP to run the DDM/HANA Centers of Excellence: generating demand for our platform solutions, specifically taking our database solutions beyond our own applications. We find new markets and solve customer problems by industry and use case, and systematically seek out customers to find benefits of the HANA platform.
It’s a fantastic time to be in the Enterprise Software business, and I’m super excited to join such a talented team at SAP.
Certainly a great success so far.
To the second part of the title "What’s Coming Next": are you also considering GPU powered data base designs to boost performance even further?
Great question. We use the Nvidia Volta GPU-based systems in our ML systems. Also see here. We more consider GPUs as a sidecar for our HANA x86 and IBM Power-based RDBMS, for certain processor-intensive activities.