OpenStack Takes a SnapShot of Its Customer Usage Patterns
Leaders of the OpenStack Foundation presented a profile of OpenStack adoption at the semi-annual Summit in Boston (May 8-10), based on a survey of 1,300 OpenStack developers and users in 78 countries. The Foundation’s key objective: Making clear that production deployments are growing, building on projects begun since 2015, even as some users are testing proof-of-concept (PoC) projects.
This blog will show key data points presented in the survey, looking at OpenStack’s challenges and the path forward for customer adoption and consumption models.
The OpenStack Foundation’s survey found that OpenStack software is having an impact in datacenters worldwide, showing 44% growth in production deployments year-over-year. In all, more than 5 million cores are in production worldwide, based on OpenStack data. More details can be found on the OpenStack website, posting details and analytics to study the survey data.
Many of the earliest OpenStack users are large companies like AT&T, eBay, General Electric, the U.S. Army, and Verizon. Organizations like these have had large and sophisticated IT staffs that have been working with OpenStack software for most of the seven years it’s been generally available to build cloud stacks.
However, the OpenStack survey showed growth in deployments among SMB organizations. More than 30% of OpenStack users work in companies with 10,000 or more employees, according to the survey – and 25% work in companies with fewer than 100 employees. The rest – 45% of respondents – include midsize companies (100-1,000 employees) and large ones (1,000-10,000 employees).
Paths to OpenStack Adoption
Customers have climbed a learning curve with OpenStack – and many are gaining operational benefits from applying OpenStack to hybrid cloud strategies, running on-prem and off-prem workloads. The next step is to extend into multi-cloud deployments, making interoperability and ease-of-use important criteria.
These customers are building cloud infrastructure based on the top OpenStack services, including the Keystone identity service, Nova compute service, Neutron networking service, Horizon dashboard service and Cinder block-storage service.
However, many organizations that are just starting with OpenStack, or electing to go with single-vendor cloud stacks suppoting OpenStack technology. These deployment patterns are often based on long-term relationships with vendors that build branded cloud infrastructure software – and those stacks may include OpenStack building-blocks.
Key challenges for OpenStack adoption include:
- Reducing the complexity and number of emerging OpenStack projects, in terms of tracking new projects and reducing minor differences between similar OpenStack community projects.
- Increasing ease of use – making it easier for first-time users to adopt the technology, and to integrate it into existing infrastructure. This is aimed at faster adoption and operational efficiency in customer sites.
- Ensuring interoperability for application development tools. Getting DevOps buy-in through OpenStack support for widely used APIs and software tools is important for wider adoption.
- Supporting deployments for OpenStack services that span multiple clouds – whether public, private or hybrid. This is why the OpenStack Foundation is working more closely with large cloud service providers (Amazon Web services, Microsoft Azure and Google Cloud Platform).
Many Consumption Models
To speed OpenStack adoption, and to address the technology challenges of working with multiple OpenStack building-blocks (e.g., Nova, Cinder), OpenStack adoption takes many forms. Mark Collier, COO of the OpenStack Foundation, said in his keynote that OpenStack is composable, open infrastructure, allowing customers to “consume different pieces of OpenStack and [to] combine them in useful ways with other open source technologies.”
Here are some of the ways OpenStack will be consumed by customers, depending on their experience levels, the time/cost of implementing OpenStack in data centers –and their budgets:
It’s already true that many organizations are electing to go with mixed-vendor or single-vendor cloud stacks that leverage OpenStack services. These adoption patterns are often based on long-term relationships with vendors that build branded cloud infrastructure – and they include OpenStack building-blocks.
For example, Red Hat, SUSE and Ubuntu were all present at the Boston conference, offering to work with customers to build cloud solutions based on OpenStack technologies – and to provide service/support for OpenStack deployments.
By engaging with named vendors, much of the work of building cloud services on OpenStack software layers is being done by others – requiring fewer IT personnel to build applications or microservices. These adoption patterns are often based on long-term relationships with vendors that provide branded cloud infrastructure – and they include OpenStack building-blocks.
Remote management for private clouds is another type of solution. OpenStack pointed to remotely managed private clouds, an emerging type of solution that is delivered via cloud service partners.
OpenStack + Containers
The move in DevOps to containers has brought OpenStack support for Kubernetes orchestration software and the Docker container software. Mix-and-match deployments reflect commercial realities. For example, Red Hat sells its OpenShift Container software product, which works with Docker, Kubernetes, and other types of open-source software. Other ISVs, including SUSE and Ubuntu, and variety of software app-dev tools providers, support combined software solutions for Kubernetes orchestration and Docker containers.
Custom solutions based on OpenStack technology are still a consumption model, especially among early adopters and large-scale IT organizations. Companies like AT&T, T-Mobile and Verizon – all telecoms suppliers – have been leveraging OpenStack to build cloud services from the data center to the “edge” of the network. These projects coincide with telco’s move to 5G wireless technology. Other companies speaking at the conference, like GE Digital HealthCare, described data center footprint savings and greater efficiency associated with their adoption of cloud computing and OpenStack-based services replacing legacy applications.
In the 2017 survey, typical customers run up to nine OpenStack services in their infrastructure, with 16% running 12 or more services. The scope of deployment is truly global: More than 85% of all clouds use some OpenStack services, and some customers have OpenStack running in up to 80% of their cloud infrastructure.
The open-source Kubernetes orchestration software is far and away the leading “application” running in OpenStack deployments (See Chart I, below).
As one customer example shows, the driver for cloud migration is efficiency and contained costs. Patrick Weeks, senior director of digital operations at GE Digital Healthcare, said OpenStack is key to moving more workloads to the cloud. Since working with OpenStack technologies, 530 applications were migrated to cloud, with 608 applications retired; in all, 42% of all of the company’s applications are running in the cloud. Using OpenStack, the GE unit achieved annual savings of $30 million, and reduced its on-prem “footprint” for systems by nearly 50%.
Chart I: Top Applications for OpenStack*
45% run Kubernetes
18% run OpenShift
18% run Cloud Foundry
17% run Custom/Build your own
14% run Mesos
14% run Docker Swarm
17% run other infrastructure applications
(*Open Answer Question, multiple responses per respondent)
Source: OpenStack Survey Data, 2017, n=1,300
Where Does OpenStack Go From Here?
Clearly, OpenStack will remain a force in cloud computing for many years to come. Support from large vendors – and wide geographic coverage in the Americas, EMEA and Asia/Pacific – say it will retain an important role in hyperscale and enterprise cloud computing.
Its technology will become an integral part of many cloud solutions – including packaged and custom code – that will make their way into the rapid expansion of cloud services worldwide. In many cases, the technology will be deployed with software from vendors, ISVs, cloud service providers (CSPs) and channel partners.
Yet, OpenStack’s importance in the marketplace is still being determined. As in earlier eras, the world will look to consistent benefits – and continuing growth in deployments – to make the case for OpenStack in public, private and hybrid clouds.