My Next Journey

An Update to all my Friends and Colleagues:

In April of 2017 - I left SAP. As a lot of you know, I wanted to take some time off to spend with my children (now 8 and 5 years old) as well as focus on getting healthy (now finally below 20% body fat). I'd be remiss if I didn't also mention all the great times I've had with my wife as we come up to our 10 year anniversary this August. All in all - a good time and the year off let me think about next steps and where I want to go next in my career.

My Discovery

Taking a year off (with the occasional spot advisory/consulting work) has given me time to reflect on the market and where I can contribute most broadly to where the industry is going. I've developed a thesis around the following points:

1. Data as the value driver - We're at the beginning of a change in what defines value in the Enterprise. Historically - applications and people were at the center of the value chain. I believe going forward, data will be the center of the value chain. We've already seen consumer companies monetize data through advertisements (a la facebook/google) and it's no surprise that some of those companies are driving effective Machine Learning and AI programs as they have data which allows building automation.
2. Application Generation is becoming possible - I met a company Decision Engines who's founder worked at some Robotic Process Automation companies (RPA) in the past. While historically, RPA processes were great at automating repetitive tasks, they were cumbersome to built and weren't easily adaptable. As data becomes available, and workflow rules make it easier to adapt, I believe a next generation RPA company with access to user clicks from applications can build rules and generate applications using Machine Learning and AI techniques. Decision Engines is headed in this direction and I've encountered a few other companies and are starting to look at this space. This, I believe, is the start of machine learning from human data which could found the basis for Artificial Intelligence to generate Enterprise Applications in the future.
3. Internet of Things (IoT) is like air and isn't high in the (monetization) value chain. I believe IoT will become pervasive. However, I don't believe in the market valuation of lots of companies in this space. I believe that right now, it's possible to build $50-$100M companies in this space, but over the long haul, all applications will need to deal with the 3 major sources of big data: User Clickstreams (facebook, web browsing, etc...), Machine Sensors (IoT), as well as Media (Pictures/Videos). The monetization will likely be on the data storage/compute/processing/analytics. Today, people don't pay a lot for drivers (storage drivers/enterprise connectors/etc...). In the same way, I believe that all applications will be connected, but monetizing IoT discretely is a fool's errand. Either build an application or have a full stack enabling technology as I described in my earlier blog.
4. Data Storage/Compute/Processing technology is far from simple at this point. The data architecture today of software is very fragmented. I believe that a modern data platform needs the following:

• Support for all object types within a single database, including Row Table, Column Table, Key-Value Pair, Graph, Document Store, File Store, etc...
• Compute Engine that is inherently multi-core and can scale-out - With concepts such as data distribution/partitioning/scale-out memory/core ratios etc... all abstracted away from a user via cloud IaaS
• Price/Performance control - Imagine a volume dial where you can turn up performance for a price without having to worry about whether the data is on a hard disk, flash, in-memory or CPU Cache
• Ability to grow - Any data platform needs to grow with the growth of data. This is where Cloud IaaS providers are definitely helping, but the software that abstracts this all away is still too complicated.

5. Machine Learning and Artificial Intelligence is at the early side of value - I fully believe in a future where Data becomes the central value point on a simple storage/compute platform with price/performance control in the hands of users where Machine Learning/AI/RPA techniques can generate Intelligent Enterprise Applications. However, I believe the biggest barrier to this today is building and managing data and making it available at a price/performance ratio that makes sense for ML/AI. This is the only way new applications can cover the entire application stack vs just the "High Value" side of the application stack. Until this problem is solved, I believe ML/AI will likely not move to main-stream in the Enterprise.

What's Next

So as I developed the above thesis, I new that I wanted to go somewhere to help solve one of the above problems. I've been working in the data and cloud domains at SAP for the last decade of my life or so, and therefore, it was natural for me to gravitate towards solving problems in this space. Interestingly enough, I got a call a few months ago from Pure Storage which is an "All-Flash" data storage company. Given that historically, storage has been disk based, and modern compute applications built on SAP HANA and other in-memory technologies have focused on the high end of price/performance (DRAM centric), I thought it interesting to explore whether the economics of All-Flash could really play a role in the storage/compute requirements of the Big Data space.

My first meeting was with Charles Giancarlo, the CEO of Pure Storage. He had a vision for their role in the future that included the more prominent role of data in the value chain. I believe he said that today, "Applications are Big and Data is small" but in the future "Data is big and Applications will be Small". A much more concise way of reinforcing my thesis. Long-story short, I decided to accept an offer as General Manager of the Flash Array Business Unit at Pure Storage and start there on June 5th, 2018 (tomorrow). Along the decision making process, I met a phenomenal team of engineers, a great management team, and one of the most enthusiastic set of customers I've seen. The customer satisfaction and net-promoter scores were off the chart. I even spoke to some customers who are using Pure Storage with SAP HANA to support next-generation Applications, which is a space that is near-and-dear to all the work I've done in the past (I'll write more about this later).

The future - The 3 legged stool

Charlie spoke to me about the the fundamentals of the hardware architecture stack today - compute/storage/networking. As he sees it, compute and networking have advanced, and now it's a time for storage to pull it's weight.

At the same time, I've been focused on making Application Development Easier. With SAP HANA, we tried to change storage/compute and move application computing closer to data in a high performance way, and with SAP Cloud Platform, the goal was to not just make the data/compute simpler, but all aspects that an application developer needs to build an application quickly. However, application development has historically dealt with the 3 legged stool of the software stack: Client/Middleware/Database. Historically, the Middleware (JAVA servers/Web Servers/ABAP Servers/.NET Servers) was where scale occurred because data platforms couldn't handle scale.

However, in a data centric architecture, we need a database/compute platform that scales where the data sits. For high value Enterprise Data, that could likely reside all in DRAM, but as we take Clickstream/Media/Sensor Data into account, scale wasn't good enough in the disk centric storage architecture. This is where I believe FlashArray from PureStorage can help balance the equation to support modern application development. This way, web scale applications using Enterprise Structured Data and Big Data (Click/Media/Sensor) can all be done simply.

Very Excited

Long story short - I'm very excited to work with a team at Pure Storage that has a great vision for the future and can help make the future data centric architecture pervasive. My goals as I join are as follows:

1. Make Application Development easier for humans
2. Allow businesses to use all data without compromise (no first class/second class syndrome for data)
3. Allow for the potential of Machine Learning / AI through enabling data/compute that scales and is manageable.

Finally, I had the pleasure of seeing many careers grow and making many friends at SAP, and I hope to do that again at Pure Storage!