Data Virtualization Expands Container Use to Enterprises
Posted in tech
By Graham Smith, Primary Data Director of Virtualization Product Management
Containers are lightweight, fast, and provide componentized environments for applications. While enterprise adoption of containers has been slow so far, the general market and enterprise interest is hot. A recent TechTarget article, Cloud market bracing for container technology showdown, notes that Docker and CoreOS are the primary players, and “dozens of new container startups [are looking] to complement container offerings, providing management and hosting environments, as well as development tools.”
One of the biggest hurdles to enterprise adoption is that there hasn’t been a way to properly manage failover and live migration of container data. As a consequence, the vast majority of container deployments are stateless, and therefore unsuitable for most enterprise use.
Container Immobility Limits Enterprise Adoption
Containers are typically limited to the resources of the server they are running on. This means that enterprise IT environments tend to do the following:
1. Limit initial use to applications with lower availability requirements. In the enterprise, disruption to business often trumps cost efficiency. Since most containers cannot be moved without downtime, early container deployments will be limited to applications that can accept downtime.
2. Excessively overprovision servers to protect against “noisy neighbors” and to reduce the need to schedule downtime to upgrade hosts.
3. Restrict usage to contain costs. Because containers are limited to host storage capacities, and host capacity costs rise quickly, organizations will tend to limit container use to applications with shared-nothing architectures.
DataSphere Enables the Enterprise to Manage Persistent Container Data
The DataSphere platform automates the movement of container data by policy and objectives that determine the container data’s performance and HA requirements. This greatly increases the number of container use cases in the enterprise, addressing the challenges described above and delivering three key benefits:
· Reduce overprovisioning and increase scalability. IT can provide containers access to a multi-tiered dataspace and set policy based objective, so fast media like flash can be used for containers hosting critical applications. They an also prioritize these applications so that “noisy neighbors” are automatically moved to other media when resource contention threatens Service Level Objectives (SLOs). Less data-intensive containers can be moved to cost-efficient capacity storage, reducing host capacity costs while allowing capacity to be scaled out easily.
· Extend container benefits through shared storage access. By making data accessible to all containers, enterprises can containerize multiple applications that require global access to persistent shared storage. Shared storage also greatly reduces host capacity costs and eliminates the need to replicate data across hosts.
· Container mobility. Shared storage enables live migration of containers when combined with technologies such as CRIU (checkpoint/restore in userspace - see below). This enables load balancing and non-disruptive maintenance for containers.
Live Migration Across the Datacenter with CRIU
A promising technology called CRIU (checkpoint/restore in userspace), announced at Fosdem 15 offers a solution for Live Migration of containers between servers. When coupled with the DataSphere platform’s ability to replicate data for high availability, enterprises achieve both high availability of persistent data and images and live migration of containers. By eliminating service disruptions when moving container data and images, enterprises can containerize many more applications than currently possible.
Setting the Containers Free
Data virtualization through automated data mobility from DataSphere frees containers from key constraints that may slow enterprise adoption. It allows organizations to protect their performance SLAs; enables a more robust and flexible shared storage architecture; cuts costs by reducing the need to overprovision and dedicate host servers to apps; and reduces maintenance overhead and preserves continuity by automating the migration of data.
Containers are an exciting innovation that promises to greatly enhance enterprise flexibility and agility if hurdles to enterprise adoption can be overcome. New technologies such as CRIU and data virtualization promise to do just that.