Real world lessons from AWS re:Invent 2015

20.10.2015
Last week Amazon Web Services (AWS) held its yearly re:Invent and it was, as always, revelatory. As I said to one fellow attendee, you go to VMWorld to see the future of old IT, and re:Invent to see the future of new IT.

If you’re interested in the future of new IT, here are three things that re:Invent made clear regarding AWS: 

Over the years, many people have denigrated AWS as a service primarily used by startups and SMBs. If ever that were true, re:Invent 2015 made clear AWS is now squarely in the mainstream of IT. 

The most obvious evidence of this was the size of the event – 19,000 attendees, up 50 percent since 2014. And the attendee look and conversational topics of hallway discussions made clear that this audience was largely made up of enterprise IT personnel. 

AWS took pains to hammer this home with keynote presentations by Capital One, GE, and John Deere, each of which vouchsafed their commitment to AWS. The GE CIO said "AWS is our trusted partner who's going to run our company for the next 140 years" and noted that it’s planning on shifting thousands of applications to AWS so that it can reduce its data center inventory from 34 to four. 

[Related: 3 Takeaways from Amazon’s re:Invent cloud conference

During his presentation, Capital One’s CIO said: “we can deploy some of our most critical workloads on Amazon.” This certainly contradicts the shibboleth that financial services companies will never use public cloud computing due to security or regulatory concerns. 

AWS also announced the creation of a joint venture with Accenture designed to help enterprise IT organizations build and operate AWS-based applications. If ever something indicates how much AWS is now mainstream, it’s the imprimatur of the premier global SI starting a company focused just on your technology (the last time Accenture did this, it was a JV with Microsoft called Avenade). 

Finally, making it painfully clear just how mainstream AWS has become, Amazon’s Andy Jassy cited its revenue. AWS is now a $7.3 billion annual business, growing at 81 percent. Just last May, Jassy said AWS was growing at 49 percent “and accelerating.” He wasn’t kidding. I wrote about AWS revenues a while ago and made some revenue projections, and I’ll likely revisit this topic; however, these numbers make obvious the fact that AWS is an enormous force in the enterprise IT space. 

Jeff Bezos famously said of retail competitors “your margin is my opportunity,” and following this dictum has laid waste to companies like Borders and Circuit City. AWS has obviously taken this perspective to heart and eyed the much fatter margins typical of the technology industry, concluding it should apply the same scorched earth approach to legacy vendors. Many of the many product it made are clearly aimed at migrating applications and data from on-premise data centers and taking share away from incumbent technology providers. 

The new QuickSight data visualization tool was explicitly announced as aimed at Cognos, although I overheard attendees also mention Tableau as potentially threatened as well. QuickSight is one of the AWS services aimed not at applications or developers, but at end users. In the case of QuickSight, the target user is a business analyst who is typically frustrated at the lengthy process required to get access to data via on-premise proprietary tools (something I’ve heard much about over the years). 

The Amazon Database Migration Service/Schema Conversion Tool might as well have been named “The Oracle/SQLServer Liberation Service,” as it aims to enable users to migrate proprietary database data into a lower-cost AWS service that provides the same stored procedures/triggers functionality. AWS probably expects a warm welcome from users for this service, given the increasingly aggressive license audits Oracle has been performing over the past couple of years. Likewise, the MariaDB RDS service offers MySQL compatibility with an open source code base not under Oracle’s control. 

Finally, AWS took aim at one of the knottiest problems facing users who would like to migrate to AWS: data gravity. Data gravity, a concept promulgated by noted industry luminary Dave McCrory, notes that data, once placed, is extremely difficult to shift due to the difficulty of network transit. That’s certainly an issue today, when organizations often have hundreds of terabytes of data located in their data centers. The new Snowball service providers users with a device (which, to my eye, resembles a picnic coffee urn) on which to place encrypted data from an on-premise location; this device is sent back to AWS via a parcel delivery service, whereupon AWS shoves the data into S3, and Bob’s your uncle. Here is a blog posting by AWS evangelist Jeff Barr that describes the service. Snowball definitely reduces the friction of migrating data into AWS where users can then operate on it with all of the AWS services. 

All of these announcements are terrible news for the incumbent technology industry. While AWS has heretofore primarily hosted greenfield applications, leaving legacy applications to be treated as cash cows by the vendors whose technology is embedded in them, now it’s directing its efforts at a takeaway game. And many users, such as those cited in the previous section, are ready to respond with alacrity. The poor financial results most legacy vendors have released over the past couple of years are likely to begin looking much, much worse. 

AWS is famous for its torrid pace of improvement and innovation and often chooses re:Invent to launch its splashiest new services. Last year it was Lambda, the previous year Kinesis, and the year before that Redshift. This year, in addition to the previously-mentioned QuickSight, Database Migration Service, and Snowball, AWS announced improvements to Kinesis, EC2 Container Service, and numerous DevOps improvements. 

The biggest announcement, to my mind, was its IoT Service, which is actually an aggregate of a number of capabilities. AWS has partnered with several chip manufacturers to provide integrated device registration and security, which reduces the burden on individual device developers. The service can securely connect devices to the backend endpoints. And it makes it easy to connect applications endpoints to other AWS services like DynamoDB, Kinesis, and Lambda. And it can support all this at scale – billions of devices and trillions of messages. Most people, even those who are IoT enthusiasts, vastly underestimate the coming deluge of connected devices and data. The IoT service clearly indicates that AWS recognizes what’s needed and is building out a cloud service to support those future requirements. 

[Related: Amazon touts improved cloud security at re:Invent

And the technology ecosystem is responding. The re:Invent program was chock-a-block with sessions describing how users are building innovative solutions on top of AWS. And these solutions are well past using AWS as a low-friction IaaS platform that delivers virtual machines quickly. 

For example, I attended one session that described how one company built out an OTT (over-the-top) video service on AWS. The application didn’t even use EC2, but instead used Lambda, the Elastic Transcoding Service, DynamoDB, S3, and other services. The kicker The company that built the service is Verizon – which offers its own cloud. The big difference, though, is that if the group building the application tried to use the Verizon cloud they would have had to install and manage all the software components and then manage scale and resiliency. By leveraging AWS, all of those services were pre-configured and ready to go, allowing the application group to focus on business logic. 

The ecosystem adjustment is even occurring in the vendor space. While at re:Invent, I met with an interesting storage startup called ClearSky, helmed by Ellen Rubin, who founded (and sold to Verizon) a cloud application migration company called Cloudswitch. ClearSky has rethought the question of where data should be located, given the scale and low-cost of AWS S3 and the latency issues of sending data across a lengthy network (this, as you’ll note, is what accounts for data gravity). 

ClearSky places a small device on premise for quick access to “hot” data, and places “cold” data up in S3. However, the fact is that cold data can become hot, or at least tepid, which requires low-latency access. So ClearSky places storage in local metro areas, where data that is likely to be needed can be placed in readiness for retrieval to the on-premise data center. Naturally, there is a ton of smart software shuttling the data back and forth among the tiers, pre-fetching data that may be needed shortly. 

The point of this is not to predict success for ClearSky, but to note that innovative technology solutions are now being developed that take the functionality, scale, and power of AWS as a foundational building block when designing new products. 

This year’s re:Invent indicates just how powerfully cloud computing is changing the technology industry. From my perspective, there are two companies setting the pace for the future of IT: AWS and Microsoft, which has shaken off a decade of somnolence to deliver a powerful and rich cloud offering in Azure. 

Both are delivering an enormous range of services, including integrated offerings that address key areas of where technology is going, like IoT, Machine Learning, and Big Data Analytics. Both AWS and Azure appear to recognize that full functionality cloud computing is the future of the industry, and whoever is the victor in this contest is poised to become the 21st century’s IBM, Oracle, SAP and Microsoft rolled into one. 

It remains to be seen how the established vendors will respond to this assault. Most have moved beyond denigration of public cloud computing in general and AWS specifically (as in this infamous example from two years ago when VMware’s COO derided AWS as something offered by a mere bookseller). The phrase of today is “hybrid cloud,” by which vendors mean a mixed infrastructure environment spanning public cloud and on-premise, and one in which they have domain expertise. 

However, the hybrid future envisioned by most legacy vendors appears to be one in which perhaps 80 percent of all applications remain on-premise and only a fifth reside in public cloud environments. In part, this reflects an assumption that legacy applications are architected and operated in methods unsuited to public cloud computing. It must be said, however, that it also reflects a rosy assumption by legacy vendors that IT organizational inertia will keep those applications rooted in environments congenial to and controlled by incumbent vendors. As the examples cited at the top of this piece indicate, that assumption may be badly misaligned with user direction. 

As I’ve stated many times, we are witnessing a sea change in the technology, one which will transform both vendors and IT organizations. For those unprepared to adapt to this sea change, the next ten years will be unpalatable and dangerous. For those who embrace it, though, the next 10 years will be glorious. The big question is: which side of the divide are you on

(www.cio.com)

Bernard Golden

Zur Startseite