A couple of weeks back at the OpenStack Summit in Vancouver, the Cloud-A team sat through a session called “Enabling Business Agility with OpenStack” with Dave Pitzely, Director, Infrastructure at Comcast, Mark Baker, Ubuntu Server and Cloud Product Manager at Canonical and Yih Leong Sun, Principal Software Engineer & Cloud Specialist at Liberty IT.
The talk was focused on how to prepare an enterprise for adopting private true cloud technology, the considerations to make, how to maximize business agility and recommendations for successful adoption of the new technology . While the session was private cloud focused, we feel that there was plenty of content that was relevant for an organization who is considering adoption public true cloud technology like Cloud-A. Here are a few point that the speakers made and how they might pertain to an enterprise looking to adopt public cloud resources.
Developer Adoption is Key
There may be cases to be made for moving all or some of your existing applications to a public cloud like Cloud-A, but the true value is in the “green field” where your developers can create modern, scalable software and bring it to market faster on the new, agile cloud environment.
Achieving buy-in from your development team is the key to a successful voyage into modern cloud technology. It is important to highlight the benefits of this technology in terms that make sense to developers. Technology like OpenStack empowers developers with the right toolsets like a self-serve model, where they are enabled to rapidly spin up their own environments on demand.
Accelerate Non-Production Activity
The agile self-serve, rapid deployment model that cloud technology like Cloud-A provides allows developers to spin up non-production environments like test, dev and QA on demand to accelerate these activities. The speakers noted that Ebay and Paypal have moved to a 100% agile cloud model for all of their test, development and QA activities. Learn how development teams are using Cloud-A for continuous integration here.
Consistent platform for Test, Dev, QA and Production
Rather than performing development and testing on internal, physical systems and launching to production on different but more powerful and resilient platform, modern cloud environments like Cloud-A empower developers to perform development, testing and QA on the same technology as production for a more consistent product. The utility billing model of public clouds also allows developers to build up and tear down these environments on demand and only pay for them when they are in use. Learn more about the cost benefits of doing test/dev/QA on Cloud-A here.
Selecting Existing Apps for Migration
The panel spoke at length about the importance of selecting the “low hanging fruit” of existing applications for migrating to a modern cloud platform like Cloud-A. Dave Pitzely of Comcast noted that while vertically scaled applications like ERP, or financial systems can be moved to a modern cloud platform, they are likely not your best bet to start. Pitzely stated that these applications tend to take time to migrate, may not perform as well as they did on their legacy environment, may have reliability issues, and the value yield will be low.
The panel recommended focusing on existing applications that are highly transactional such as data analytics, event mediation or applications with advertisement and click through data. These applications are highly variable and highly scalable and lend themselves to the API driven infrastructure of a modern cloud platform, thus producing a high yield of value. Learn more about how the application “Lift and Shift” to the cloud technique fails and some best practises for refactoring your apps for the cloud here.
Differences in App Architecture
In the talk, Yih Leong Sun noted the differences in traditionally architected applications and “cloud-aware” applications. Sun noted that traditional applications tend to be monolithic, centralized, have tightly coupled components, are synchronous and typically single tenancy, whereas “cloud-aware” applications have distributed microservices, are asynchronous, multi-tenant, have built-in failure-resistance, have decoupled components and are eventually consistent.
The characteristics of “cloud-aware” applications gives them functionality that traditional applications simply cannot compete with. One example of this is “Cloud-aware” applications can be architected with redundancies and resilience built into them and leverage the APIs provided by the modern cloud infrastructure, whereas traditional, monolithic applications tend to rely on extremely expensive hardware for all of their reliability and uptime. learn about existing DevOps tools that help automate deployments and management of modern cloud applications here.
Handling Internal Politics
As with any department in a large organization, IT will have hurdles to overcome when making a change like moving to a modern cloud platform like a private OpenStack deployment. While this discussion was focused on private cloud deployments, organizations will likely also come up against internal resistance to adopting public cloud technology like Cloud-A
The way that budgets are determined and allocated for funding a project in the public cloud is vastly different than budgeting for internal infrastructure. Because of this, organizations may come up against resistance from their purchasing teams. Utility billing, while very convenient, in most cases less expensive, is difficult to predict, and there is less ability for purchasing teams to negotiate pricing with a public cloud vendor as they would with a hardware reseller. In addition to all of this, purchasing departments are also accustomed to up front fees hardware and licensing fees, and long, drawn out RFP cycles. More about writing an RFQ/RFP to procure public cloud infrastructure here.
The other problem that modern cloud technology can create with the financial decision makers in an organization is the over provisioning of infrastructure. Because infrastructure can be rapidly deployed on demand on a cloud platform like Cloud-A, sometimes developers feel empowered to deploy an unnecessary amount of resources, which can be expensive. The panel recommended either a bill back model, where an internal department is billed for their infrastructure, or a show back model, where an internal department has regular visibility of their usage, both which can make individual departments in an organization accountable for the infrastructure that they deploy.
The panel stressed getting buy-in from the C-level prior to taking on a new, disruptive project by being transparent and highlighting and measuring the business value of modern cloud technology like business agility, innovation and faster time to market, which often times, at scale, can far outweigh the costs associated with cloud infrastructure.
Moving Forward with Cloud-A
Over the next several months Cloud-A is focusing on enhancing our user experience to enable development team adoption of our modern cloud infrastructure. We have been working with existing users, putting on focus groups and releasing surveys to help understand the the path of least resistance for enterprises adopting our technology. Stay tuned for some exciting announcements.
Ubuntu Server and Cloud Product Manager, Canonical
Mark Baker is the Ubuntu server and cloud product manager at Canonical where he has spent the last 4 years helping drive the platform for next generation application delivery. Prior to Canonical Mark worked at MySQL and Red Hat where he enjoyed disrupting large Billion dollar incumbent technology companies. With OpenStack and Ubuntu Mark continues to have fun following this same path.
Director, Infrastructure Architecture, Comcast
As Director of Architecture for Comcast, Dave oversees the corporation’s strategies for distributed private cloud, hardware, application design and data persistence platforms. In this role, he also supports its foundational infrastructure for national data centers and distributed applications, including such areas as Cloud Computing, IT Governance, Storage, Hardware, RDBMS/NoSQL, Application Platforms and Virtualization. Prior to… Read More ?
PhD Cloud Computing, Liberty IT
Leong accumulated 15 years of experience in software development and infrastructure deployment. He obtained PhD Computer Science (Multi-Cloud Infrastructure) in 2013. He spent the past 6 years on Multi-Cloud infrastructure development. He is currently a Principal Software Engineer for Liberty Mutual Insurance Group, working for the next-generation Cloud platform project. Prior to that, he led the engineering team of a few start-up companies in… Read More ?