When the phrase "hybrid cloud" is mentioned, you might get different reactions from people or variations on the definition. But as we enter 2016, I believe the hybrid cloud is real and that it is going to continue to shape cloud computing. And if you trust the industry research, this multi-billion dollar industry is growing quickly as organizations continue to expand their data center and cloud strategies.
To dig in deeper, I recently spoke with an industry expert in the field of hybrid cloud technology, Andrew Hillier, who also happens to be the CTO and co-founder of Cirba, one of the leading providers of software-defined infrastructure control solutions. I wanted to find out, is Hybrid Cloud right for everyone? If a company goes down that path, what's needed? What are the benefits? And what happens next?
VMblog: We are hearing more and more about hybrid clouds these days. Wanted to get your expert opinion on the subject and ask, is hybrid cloud right for all businesses?
Andrew Hillier: I think every business can benefit from the use of public
cloud in some form, even if just for a handful of temporary workloads.
That may not be considered true hybrid cloud, but it still requires many
of the same management and governance structures to be in place. In
my view true hybrid cloud is when an organization has apps that straddle on and
off prem, or apps whose components can run either on or off based on loads and
business cycles. In other words, apps that can systematically leverage public
resources in order to optimize cost and fit for purpose. I believe that
most large enterprises with complex IT environments will begin to incorporate
public cloud resources into their portfolio of hosting assets, but the
challenge lies in doing so wisely.
VMblog: And what are the main challenges in adopting a hybrid cloud
strategy?
Hillier: I have heard the conversation evolve at our customers from
"we'll never do that" to "we need to figure out the smart way to do it", so the
attitude from internal IT that used to be a huge barrier to hybrid cloud
adoption are changing. But there are still very real concerns about
security, cost, and just simply trying to figure out what should go into the
public cloud. Unfortunately, organizations are currently unable to figure this
out and rely on a lot of manual effort to make the decision, which leads to
uncertainty and subjectivity in the process. They don't have a way to
objectively analyze workloads and automatically determine what can and should
be placed internally and externally.
VMblog: How can organizations automatically determine the best
execution venue for applications in hybrid cloud?
Hillier: I love the notion of the "best execution venue" because it
puts the emphasis on the individual application requirements. Having a means to
capture the detailed app requirements, everything from utilization requirements
to technical resource requirements, applicable compliance, security, regularity
requirements is the first step. You also then need to model of capabilities of
available hosting environments, both internal and external, and in a similar
level of detail. Only then can you really determine where the workloads
should go, where there is space, and how to place workloads to make the best
possible use of infrastructure. This requires analytics that use policies
to govern those decisions, automatically determining what goes where in order
to balance demand and optimize the cost / benefit impact.
VMblog: Explain if you would, what are the main benefits of having analytics-driven
decisions of where applications should be hosted?
Hillier: Some people ask us for opinions of whether an app can go
into public cloud. It isn't a matter of opinion, it's a matter of whether
or not the environment is "fit for purpose", whether it has the required
resources, and whether it is the most cost-effective option. There are
lots of things that can go wrong by placing certain types of workloads out into
public cloud - security risks, compliance risks, performance, etc..
Having analytics to govern those placement decisions prevent those
situations from occurring, while at the same time optimizing cost. And
cost is a whole topic on its own - there is a notion that public cloud is
always cheaper than on-premise hosting and that just isn't the case. We
have done analyses of production environments and found that, depending on the
nature of the workloads, there are significant variations in the costs of
public vendors, bare metal options and even on-premise hosting. Analytics
gives you the opportunity to scientifically factor cost into the decision to
ensure that you make the best use of internal and external resources, and this
isn't just a simple calculation that can be done manually. Analytics also
enable automation of these decisions, and can seamlessly integrate to cloud
management platforms and orchestration tools to make it happen.
VMblog: Don't OpenStack or products like VMware vRealize Automation automate
these kinds of decisions?
Hillier: The short answer is no. OpenStack offers some very basic
filtering logic and VMware provides placements based on round robin decisions
and some manual checks. The strength of those solutions is in the
automation mechanisms they provide, and they are very good at provisioning VMs,
but not necessarily at determining where they should go. This is why we
complement them very well, and Cirba integrates with both these tools to plug
in new decision logic, enabling an intelligent, automated process for real time
routing and placements. Some of our customers run multiple stacks, and
have both OpenStack and VMware as hosting options. In these cases Cirba
sits on top of the cloud management layer, providing a higher level decision of
which technology platform a workload should go into, then which environment
within the chosen platform and finally which server level placement. These are
some advanced use cases, but if you think about it this scenario is really the
same decision making that is required for hybrid cloud adoption.
VMblog: Once you put them there, how important is it to actively manage the
workloads operating in public cloud environments?
Hillier: It's extremely important to properly manage all your
workloads, whether on or off-premise. When running on internal
infrastructure, its critical to have systems in place that can ensure workloads
get the resources they need to perform as required, while at the same time
keeping unit costs as low as possible. This means managing VM density and
placing workloads strategically in order to minimize overall cost, including
software licensing requirements. For public cloud, the emphasis becomes
allocations, which must be watched like a hawk, or you'll overpay for capacity
that you're not using. It's also important to be able to know whether it makes
sense to repatriate apps and bring them back in house, such as after a peak
business cycle has passed. And if you are leveraging public bare metal, you
have to manage those environments in exactly the same way you would for
on-premise environments.
VMblog: Is moving to hybrid cloud primarily a technology problem?
Hillier: I think people underestimate the governance and financial
aspects of the transition. There is a lot of focus on the plumbing,
automation and the "how" of hybrid cloud, but the "why" is extremely
important. Policies that govern why apps can and should go to certain
hosting environments are key, and the whole decision making process must be
codified and automated. So fundamentally it's a business decision problem that
requires analytics to solve, and it must be done fast enough to allow IT groups
to meet the dynamic needs of the business. If IT can't make smart decisions
about workload placements that the business trusts, and do so quickly, then the
business will take matters into their own hands, which usually isn't in
anyone's best interests. We have seen organizations face this challenge just
trying to get internal clouds adopted, and if they don't resolve that
fundamental issue before they move to hybrid, they will have an even bigger
mess on their hands.
VMblog: Finally, tell us what Cirba does to help organizations adopt
hybrid.
Hillier: Cirba is analytics software that provides workload routing
and granular control over hybrid cloud environments that include
Microsoft® AzureTM,
Amazon Web Services
(AWS) and IBM®
SoftLayer®, VMware®, IBM® AIX on PowerVM, KVM,
Red Hat® Enterprise Virtualization and Microsoft® Hyper-V®.
Cirba provides the necessary decision control and governance
for automating the decision of where applications can safely run, across both
internal and external hosting technologies. Once there, Cirba provides
centralized management for enterprise applications across these hybrid cloud
environments by giving customers visibility into where applications are hosted,
whether they are appropriately resourced, whether they are running efficiently,
and whether they are better off moving somewhere else. To do this,
Cirba's analytics capture the detailed application requirements such as compute,
storage and network requirements, security considerations, compliance
requirements, software licensing requirements and other operational policies
and matches these to the capabilities of available infrastructures. Coming this
year, we are adding hybrid routing support to our routing capabilities
currently available for private cloud.
##
I'd like to thank Andrew Hillier, CTO and Co-Founder of Cirba, for taking time out to speak with VMblog and answer a few questions about the company and Hybrid Cloud.