|By Dave Jilk||
|January 4, 2012 08:00 AM EST||
Despite my shaky prediction performance in 2011, I'm going to try again.
Last year, my best predictions were based on some sort of specialized knowledge or insight, and the worst came from looking into the macro environment and making guesses. So, this year's predictions emphasize the former approach, yet hopefully avoid sounding like claims such as "the market will fluctuate."
1. 2012 will be the Year of the Storefront in the cloud.
Admittedly, this is a somewhat self-serving prediction, since Standing Cloud is (among other things) a storefront. But based on the variety of offerings, rumors, customer behaviors, and other sources I have, it seems that the confluence of forces that helped drive the popularity of App Stores in the mobile world are now propelling the storefront concept ahead in the cloud world.
Viewed through a relatively simple lens, a Storefront has three basic components:
- A catalog of offerings (applications, utilities, services) with information about each, such as features and benefits, user reviews and ratings, licensing and pricing.
- The ability to purchase or obtain access to the offering.
- The ability to easily install or deploy the offering.
These issues are all more complicated in the cloud (whether private or public). Unlike a mobile app, deployment of the offering is not necessarily on a single device; they must be customized or integrated with other offerings, and pricing may have many components.
Nevertheless, the same need for informed and easy procurement exists in the cloud and will drive development and adoption of storefronts.
2. There will be a noticeable amount of customer churn from Amazon Web Services among startups that are scaling.
AWS has become the de facto standard data center choice for startups, particularly those with venture funding. Amazon even has a business development team that builds relationships with venture capital firms. Zynga, which is generally believed to be AWS's largest customer, demonstrates that a company can scale massively with Amazon.
Why do I predict that startups with fast-growing businesses and utilization will leave?
The detailed reason will vary for each company, but at a high level it will be due to the particular architecture AWS has for its network, servers, virtualization, storage, etc., and the vision that systems architecture implies for the types of application architectures that will be deployed on it.
Increasingly, I see startups that have either built application architectures that scale in a way that is incompatible with the AWS vision, or perform functions for which the optimal application architecture is incompatible with the vision.
In the first case, the scaling startup is faced with either changing its application architecture, changing its provider, or both - so it evaluates options outside of Amazon. In the second case, it may take some time to realize the incompatibility, but once realized, customers will evaluate both hybrid and alternative IaaS providers to find the best solution.
In both cases, this evaluation was likely something they skipped at the beginning due to both Amazon's status as the default provider, and because issues of scale were not yet a factor. Once evaluations are performed, some percentage of companies will switch, and providers such as Rackspace, SoftLayer, OpSource, and ViaWest will be the likely beneficiaries.
3. Cloud IDEs (Interactive Development Environments) for developers will grow in popularity
One of the most popular use cases for IaaS is development and test, but this is primarily for servers. Most developers still code on their local workstation and push code to a repository when it's ready to commit or needs to be tested. There are numerous advantages to a fully "cloud" IDE (one where the source files are always remote), including a tighter and more automated/integrated dev-and-test loop, ease of working on multiple devices and keeping development environments up-to-date, ease of adding or removing development team members, and others.
There are two primary impediments to the adoption of a cloud IDE: (1) performance and reliability of the Internet connection and (2) IDE feature set, including integration with third-party tools. Given that a number of companies are working on cloud IDEs, the latter is likely to improve during 2012 and encourage growth.
As to the former, it is increasingly the case that companies and individuals have alternative means of Internet access, all of which perform well enough to support an IDE. For example, mobile hotspots, coffee shop Wi-Fi, home broadband, and redundant Internet providers all make it less likely that developers will be unproductive for any extended period of time due to a connectivity outage.
To be clear, I am not predicting that cloud IDEs will take over the world in 2012, just that there will be noticeably more usage and buzz about them.
4. CloudFoundry will consolidate its status as the standard way to operate a Resource PaaS.
A Resource PaaS is a service that enables multiple application tenants to share granular computing resources for scalability, reliability, and ease of deployment (for more on Resource PaaS, see my previous article on multi-tenancy in PaaS). Cloud Foundry is an open source software project, and is already in widespread production use among Resource PaaS providers as well as internal projects.
Because it is a kind of meta-application server, it is becoming increasingly difficult to justify building these core PaaS capabilities in a proprietary way. Instead, Resource PaaS providers will need to dedicate their efforts to tuning their stacks and environments for particular categories of developers, for example, those using a particular language or application.
- Different Approaches to Databases in PaaS: A Round-Up
- Keys to the PaaSing Game: Multi-Tenancy
- Ahead in the Cloud: 2012 Cloud Computing Predictions
- Web Application Lifecycle Maintenance
- Cloud-Prem Software: New App Deployment Model Blends Best of Both Worlds
- The Cloud’s Little Secret
- The Transition from Virtual Appliances to Model-Driven Deployment