Still confused and searching for information about public vs. private cloud? Is your organization considering the move to public cloud but still worried about the challenges of doing so? You aren't alone. I recently spoke with Chuck Dubuque, VP Product Marketing,
at Tintri to get some answers.
Tintri, if you aren't familiar with them, offers an enterprise cloud infrastructure that's built on a public cloud-like web services architecture and a set of RESTful APIs. Organizations use Tintri all-flash storage with scale-out and automation
as a foundation for their own clouds-to build agile development
environments for cloud native applications and to run mission critical
enterprise applications. Dubuque told VMblog that Tintri is continuing to extend its enterprise cloud platform over the year ahead.
I found this stuff to be really interesting and useful -- their cloud infrastructure is designed to deliver the same agility as the public cloud, but in the enterprise data center.
trends are you seeing with the adoption of public vs. private cloud?
Chuck Dubuque: There's a myth that all workloads are
moving to public cloud, but the trend is more subtle. Industry data shows that
while traditional IT spending is declining and public cloud spending is
increasing, private cloud is growing most rapidly.
We have seen over just the last 12-18
months a real change in cloud strategy, and it is based on the real experience
of enterprise customers. A few years ago when CIOs publicly declared that they
were ‘all-in' on cloud, it was generally assumed that they were referring to
public cloud. But organizations have encountered challenges with particular
workloads, especially after trying wholesale "lift and shift" migration to
public cloud. Now we are seeing a more informed "multi-cloud" strategy where going
‘all-in' on cloud means finding the right mix of public cloud, private cloud,
and cloud service providers.
VMblog: And what are some of the biggest challenges organizations are encountering as they move to
Dubuque: We recently convened our Customer
Advisory Board-representing a broad range of organizations. We asked them
directly what kind of challenges they had encountered with public cloud and
each had a horror story of sorts. Those stories centered on predictability of
performance and cost. For example, one customer talked about a workload that
was much more network intensive than they had realized. This application's use
of network resources inside their datacenter was a hidden cost-something they
didn't have to pay for directly, but in public cloud the cost of that network
bandwidth ran up a surprise $50,000 monthly charge above what they had budgeted
based on compute and storage alone. Other customers talked about performance;
select, mission-critical workloads that performed consistently in their data
center struggled in public cloud because they weren't designed to run
effectively on public cloud, which relies on applications to provide their own
availability and resiliency.
VMblog: When should an organization consider investing more in private cloud?
Dubuque: Our position is that there are some
workloads that can safely be deployed public cloud, including many newer "cloud
native" applications. Others need to reside in an organization's data center,
whether it is for compliance, cost control or predictable performance.
When an organization has a
significant number of those workloads, they need to be investing in their
private cloud-and we believe the workloads that reside there should benefit
from the same agility of public cloud. That's the promise of Tintri enterprise
VMblog: Can you talk about the defining characteristics of enterprise cloud?
Dubuque: We look to the NIST definition of
enterprise cloud. The pillars include resource pooling, elasticity, measurement
and self-service. These are the same elements that define our product vision
Tintri enterprise cloud works with
compute and network resources to provide the infrastructure base of a private
cloud. Tintri enterprise cloud provides a single, federated pool of storage. It
can be elastically scaled by adding expanding capacity or adding nodes-and
automatically recommends the optimal location of every virtual machine. Tintri
analytics predict exact needs for capacity and performance while enabling
what-if modeling. And self-service includes managing your Tintri estate through
technologies like Slack and Amazon's Alexa agent. Our roadmap pushes things
VMblog: Explain if you would, how
is Tintri enterprise cloud different from similar solutions touted by all-flash
and/or hyperconverged vendors?
Dubuque: Tintri enterprise cloud is
distinguished by its web services architecture. That allows Tintri to work like
a building block that can easily connect with compute and network. Tintri
analytics span the entire infrastructure and can predict and control storage
and non-storage resources like compute.
Other all-flash players are
generally block storage, and they still use LUNs and volumes. That's
impractical for enterprise cloud, because it's a different language than
compute and network speak. You still require dedicated storage specialists
while the rest of your datacenter staff are becoming generalists, and you still
need manual intervention to complete storage tasks which impedes automation and
And while hyperconverged brings
together compute and storage, it also forces you to scale them both together;
but the elasticity of enterprise cloud requires that they be scaled separately
according to need. With Tintri, scaling is as simple as expanding capacity or
adding nodes-then Tintri automatically recommends the optimal location for
every virtual machine.
are some of the differentiated capabilities of Tintri enterprise cloud?
Dubuque: Tintri's architecture allows for
autonomous operation-meaning that Tintri doesn't just automate-away problems,
it avoids and eliminates complex tasks altogether. By using the virtualized or
containerized application as the abstraction layer at which everything from IO
isolation to performance and automation works, Tintri is able to manage almost
all day-to-day operation without administrative intervention. Every virtual
machine or container gets its own lane. There's never conflict over resources
and never any need for manual intervention (for example, to shuffle VMs between
LUNs or volumes). When the infrastructure is self-managing and self-healing, it
is possible to very simply add automation to create value-added capabilities. Tintri's
set of REST APIs allows the automation common tasks like replication, cloning,
policy management and more, while Tintri autonomous operation provides
predictable high performance.
The level of abstraction at which
Tintri operates also means we can do some pretty cool stuff that's not possible
with other products. For example, you can manage your Tintri enterprise cloud
through Slack or Amazon's Alexa agent. Why does that matter? It means you can
enable self-service beyond admins, empowering non-experts to take ownership of
their footprint. You wouldn't want a Developer to be carving up LUNs or tuning low-level
storage settings, but they could absolutely provision the applications they
need through Slack and Tintri.
VMblog: Finally, how
are Tintri customers deploying enterprise cloud today?
Dubuque: My favorite customer example is a
technology company that is a long time Tintri customer. They currently have 57
Tintri VMstores in operation-a footprint that spans 4 petabytes of storage
across 8 global data centers. But the truly impressive part is that they use
Tintri as an enterprise cloud to spin up and tear down more than 120,000
virtual machines every month while running nearly 200,000 in steady-state. And
they do it ALL with a single IT generalist who spends only four hours a week
thinking about storage management and growth, not fixing issues and managing
storage. Tintri's autonomous operation and powerful, scalable automation free the
rest of the organization from managing storage, so they can concentrate on
developing new ideas and products.
Thank you again to Chuck Dubuque, VP Product Marketing at Tintri, for taking time out to answer questions for VMblog readers.