The moves will enable IT departments to provide persistent storage to support containers that can guarantee key performance metrics per container.
Tintri supports VMware, Microsoft Hyper-V, Red Hat Enterprise Virtualization and Citrix Xen virtualisation platforms on its arrays and can support them all in a scale-out pool.
Its latest moves will allow customers to support containers on Tintri arrays, either via a VMware virtualisation layer or on bare metal (although there will actually be a server operating system present) via a Flocker abstraction layer.
First available will be support for vSphere Integrated Containers. This will be available when VMware releases version 6.5 of vSphere, probably later this year.
Then, in the first half of 2017, Tintri will support Docker containers via the open source data volume manager Flocker. This will support Docker containers on without a virtualisation hypervisor and inside VMs.
Tintri marketing VP Chuck Dubuque said: “Flocker provides a storage abstraction layer that presents as an NFS storage device to Flocker drivers. APIs connect to the management layer, so Tintri storage is aware which I/Os are associated with which storage objects.”
Tintri container storage support will allow customers to measure storage performance metrics on a per-container basis and to set performance requirements per container.
Containers have gained prominence recently, and are a way of virtualising applications, essentially isolating them from other applications directly on the operating system.
Read more on storage for containers
- Docker, by default, doesn’t come with persistent storage, which presents an issue to some workloads that customers want to run in containers. However, there are ways to achieve persistent storage.
- Container technologies, especially Docker containers, continue to gain traction in the storage market. This guide explores how containers stack up against virtual machines as well as who is using them.
Run like that, they are a potential alternative to server virtualisation with a more lightweight footprint. Originally, it was not possible to provide persistent storage for containers using this approach, but this is beginning to change with moves like Tintri’s.
Containers can also be run inside a hypervisor. This has gained popularity because of the ability to integrate containers more easily with existing infrastructure and services.
Containers originally did not allow for persistent storage, and this was fine for apps with a very short lifecycle where data was not critical, such as in some web environments. But there has been a rise in the use of containers for applications with longer lifecycles and these have required persistent storage.
Such is the rise in demand for persistent storage for containers that there has been a flurry of releases by storage suppliers in recent months.
Dubuque said Tintri was responding to demand from customers that use containers in a variety of ways.
“Part of it is shadow IT where an admin gets a request for one large VM and datastore, then the developers carve out, say, 100 containers,” he said. “This is an increasingly common way for DevOps to work, so what we are doing allows IT to keep some kind of control over that.
“We are also starting to see a little bit of containerisation of apps that resisted virtualisation, where even a small amount of latency is too much, such as in nanosecond timestamps in financial services or real-time timestamps in internet of things applications.”