The topic of how OpenStack should relate to Amazon Web Services (AWS) has been a source of furious debate in the OpenStack community, with some advocating direct mano-a-mano competition and others pushing for capitulation and finding a way to work with AWS in some manner. Others in the community want to explore a middle way that will allow OpenStack to flourish both as a viable alternative to as well as an uneasy partner with Amazon. My contribution to this discussion is to ask that we take a step back and look at what I believe are several viable approaches for how OpenStack can respond to AWS in today’s market. To help illustrate these approaches, we’ll take a look at their analogs in the car rental industry.
Avis Car Rental: “We Try Harder”
The first approach, of course, is to go head to head with the market leader and try to match them feature for feature. It’s a similar approach taken by Avis with Hertz in the car rental industry and Burger King with McDonald’s in the fast food industry – watch what your competitor, who is the industry leader, is doing and mimic it as best as you can, but with some added value bonuses, such as better service and/or lower prices; then over time, try to beat the leader in terms of rolling out new functionality and offerings. Google and Microsoft seems to be taking that same approach against AWS with their public cloud offerings and there are some in the OpenStack community who want to do the same. This group believes that by leveraging the scale of an open source community, it is possible to catch up to and eventually surpass AWS in functionality and features.
This approach, however, is potentially the riskiest to take in a market where you are wrestling with an industry’s 800 lb gorilla. As I’ve argued in an earlier “Which Amazon Are You Chasing” post, it is extremely difficult to try to achieve technological parity with AWS given their head-start and the number of developers they employ. I said at the time and still believe that an open source technology, such as OpenStack, has the potential to achieve this parity but it will not be easy. And along the way, the OpenStack community cannot lose sight that building a stable and usable Cloud platform is job one and the foundation for creating additional services.
Thrifty Car Rental: “Book Smarter”
A second approach is to attack an adjacent market or choose a segment of the market that you believe you can serve better than your competitor. For example, smaller car rental players like Thrifty or Budget are not necessarily looking to overtake Hertz; but by focusing on the specific value proposition of trading frills for lower cost, they can carve out a segment of the market for themselves. Likewise an OpenStack cloud vendor could decide that they will focus specifically on mid-market and small to medium size businesses that may require more hand-holding than they would receive from AWS. Many would say that has historically being the approach my employer, Rackspace, has taken by bundling their “Fanatical Support” with their OpenStack powered offerings. But other smaller public cloud providers are taking the same approach. The question is can an OpenStack provider that sticks with this approach co-exist peacefully with AWS or will the 800 lb gorilla eventually take over their market segment as well.
Zipcar Car Sharing: “Wheels When You Want Them”
A third approach is to take a concept popularized by a market pioneer/leader and adapt it to create a new market or to go after an under-served market. Zipcar disrupted the car rental industry by offering car sharing, which gives consumers the benefits of car rental without the hassle and at a lower price point. Zipcar customers pay a monthly subscription which gives them on-demand, self-service access to a fleet of cars for a pre-set period of time. Most OpenStack vendors haven taken a similar approach by building private cloud products and services using the OpenStack code. This approach assumes that consumers are looking for a private cloud alternative to AWS, particularly in the enterprise market and that an OpenStack powered Private Cloud will provide them with enough value that consumers are willing to accept the tradeoff of being on a Cloud platform that do not have the same range of services as AWS.
Avis Budget Group: “It’s Your Space”
The last approach I want to talk about is a hybrid (no pun intended) approach that blends together several offerings, each individually less robust than that offered by the market leader, but taken together creates something that offers differentiated services to the consumer. When the Avis Budget Group, the parent company of Avis Car Rental, purchased Zipcar last year, they acquired an offering that was missing from their portfolio and competed directly against their core offering. Since then, Avis has integrated elements of the two businesses to give consumers more options on how they can use a temporary vehicle. This includes fleet sharing so that the Zipcar unit can “burst up” and offer cars from the Avis unit when peak demand exceeds steady state capacity; the Zipcar unit also allows customers to choose a car from the Avis fleet if available Zipcars are not an appropriate fit for a particular customer. This “hybrid” approach has been appropriated by several OpenStack vendors who generally attack it in one of two ways:
- The first approach assumes AWS has already won the public cloud wars and that OpenStack or some other Cloud platform should focus on being the private cloud “clone” to AWS. Companies like CloudScaling and Eucalyptus seem to have adopted this approach and actively advocate for full API compatibility with AWS so that their customers can more easily move workloads between their private clouds and AWS.
- The second approaches assumes there is room for a smaller public cloud that is built on OpenStack which then can be integrated with an OpenStack powered private cloud to create a hybrid OpenStack cloud. My employer, Rackspace, and HP are two vendors who have taken this approach of building out OpenStack powered public and private clouds and enabling workload mobility between their respective private and public clouds. In particular, Rackspace has moved from a perceived focus on being a smaller version of AWS to offering differentiated services through their Hybrid Cloud offering.
The key to this model, whichever approach is taken, is to deliver a service that is able to reach the largest number of consumers possible and to be able to satisfy the broadest set of use cases possible.
When it comes to developing a model for doing business in the shadow of AWS, it is not an either/or proposition. Having a clear market leader should not preclude others from entering the market either to disrupt or to offer an alternative offering. After all, would Linux have the prominent place it has in the data center if the open source community had ceded the operating system market to Microsoft and SUN a decade ago? At the end of the day, it comes down to understanding your target market, your own capabilities, and being able to offer something that produces value for consumers. Please let me know what you think. Are there other models that vendors in the OpenStack community can adopt? Is there actually only one model that really will work while doing business in the shadow of AWS?