VMblog recently had the opportunity to sit down and speak with John Gentry, Chief Technology
Officer, Virtual Instruments. We talked about the company, and Gentry shared his industry expertise with us by explaining the current state of artificial intelligence, machine learning and AIOps, as well as how his company is approaching
these technologies.
VMblog: To kick things off, please tell us a bit about yourself, your
background and your role at Virtual Instruments.
John Gentry: I'm currently the CTO and SVP of Business
Development at Virtual Instruments, a hybrid IT infrastructure management and
AIOps company that focuses on optimizing the infrastructure supporting
enterprises' business-critical applications. Our goal is to make it possible
for enterprises to gain an end-to-end view of their entire hybrid IT
infrastructure in the context of the application.
In my current role, I'm responsible for
keeping up with and understanding key IT industry trends that affect product
strategy and strategic alliances. I originally got involved in this space to
solve customers' problems, and I started working with Virtual Instruments
because I saw that they had the best data, the best analytics and the best
technology that could solve real problems in IT operations, specifically from
an infrastructure management perspective. The problem Virtual Instruments was
solving was a problem that companies I previously worked with had struggled to
manage - their solution was to throw more hardware and more people at the
issue, which only made it more complex and resulted in finger-pointing across
the organization.
VMblog: Artificial Intelligence and Machine
Learning have started to dominate the conversation around the digital
transformation of IT operations, but many organizations are understandably
still determining how to apply these concepts in practice. So where are we in
deploying these kinds of smart technologies in IT?
Gentry: Most large enterprises are typically using
dozens of management and monitoring tools - a combination of application performance
monitoring, infrastructure performance monitoring, network performance monitoring,
and an abundance of other siloed, vendor specific infrastructure monitoring
tools. For the most part, the service management and change management tools
work well, but the monitoring side is a major problem as the tools don't have a
shared or common context and lack any awareness of the applications - they essentially
speak different languages.
Some organizations have implemented first-generation
dedicated AIOps tools that simply collect and analyze alerts or analyze logs
from all these other products. These alert and log aggregation tools are
helpful with troubleshooting, but they are all "post-facto" solutions that
can't be used for real-time performance monitoring or for proactive problem
prevention.
VMblog: In your opinion, what are the biggest
misconceptions around technologies like AI and ML?
Gentry: The biggest misconception I see around
these kinds of "smart" technologies is that many organizations see them as an
easy button that can solve all of the problems associated with managing their
increasingly complex IT infrastructure. But the reality is that, while some
AI-based solutions have promised to make sense of the onslaught of data produced
by modern businesses, you can't just layer basic correlation and de-duplication
on top of existing legacy tools and have an effective AIOps solution.
When it comes to working with AI in complex
IT infrastructure environments, it isn't just about throwing various algorithms
and data at the problem. To get to a truly value-added application of AI or ML,
you must understand the problem you are trying to solve, the context, and the
right approach to applying AI/ML to derive specific outcomes. You must have
"applied intelligence", sometimes called tribal knowledge, that provides a
knowledge base of historical experiences that can then be applied to current
issues. The combination of applied intelligence plus a specific combination of ML,
statistical analysis and heuristics, is what second-generation AIOps platforms
will include.
VMblog: Let's go a bit deeper into AIOps - I've
seen the phrase thrown around quite a bit by a number of different companies,
but what does it really mean? And how does it deliver value in the enterprise?
Gentry: AIOps is a relatively new industry term
that's increasingly gaining attention across the enterprise. It's described by
Gartner as having two main components: big data and machine learning. True
AIOps offers a plethora of benefits for the enterprise, including providing a
real-time, end-to-end view of how a business' critical applications are
performing and how the underlying IT infrastructure supporting them can be
optimized. It provides a basis for both reactive and proactive problem
identification and resolution. The key elements to AIOps include:
- Data ingestion: In order to make
better decisions, you need the best data and applied analytics. AIOps
solutions need to be able to ingest massive amounts of data from a myriad of
sources. Start at the source with data
from individual elements of the infrastructure from AIX to AWS, systems of
record from service management solutions, business context derived from
application performance monitoring solutions, all with a real time
architecture.
- Correlation: By bringing together
all the infrastructure monitoring data with service desk and application
performance monitors, true AIOps incorporates correlation across all domains to
determine how to optimize application performance based on business importance
and when there is an issue, the ability to determine what really is the root
cause.
- Visualization: Comprehensive AIOps
should provide easily understood dashboards and reports on the business'
infrastructure that can be customized to any manager's requirements - from a
simple red, yellow, green management dashboard to a full deep dive of
components, IO metrics and capacity.
- AI-Based Machine-Learning: It goes
without saying that true AIOps should feature next generation "smart"
technologies, including AI-based machine learning. This is the only way these
solutions can understand normal activity patterns and ensure you are only
alerted when any element trends outside its normal activity. If you're
constantly getting false alerts, your AIOps is not, in fact, true AIOps.
- Automation: In addition to recognizing
abnormal activity, AIOps should be able to recommend how your IT infrastructure
can be optimized and suggest courses of action to assure and improve
application performance. IT should be able to proactively recommend changes to avoid
future problems.
VMblog: How is Virtual Instruments approaching
this technology? What differentiates you from others touting their AIOps
capabilities?
Gentry: What makes our VirtualWisdom solution
different is the fact that our team is applying, quite literally, hundreds of people-years
of experience to our approach to AI, ML and AIOps. By monitoring the entire hybrid
infrastructure, we're able to generate our own unique data, as opposed to just
pulling existing data sets from various external sources. In other words, we're
not just aggregating data - we're generating and collecting it ourselves in
real-time. This data is then used to educate the WisdomAI engine, allowing the
intelligent algorithms to be fully informed and operational, rather than
glorified pattern matchers. In addition, VirtualWisdom is an application-centric
AIOps monitoring and automation platform. It knows where applications live on
the infrastructure at any given time, knows the relative business value of the
applications and how they stress the infrastructure.
VMblog: AIOps is perceived as the prerequisite
for true IT automation. How do you see the relationship between AIOps and
automated IT Operations?
Gentry: Just like any nascent industry, smart
technologies like AI and machine learning are going to continue to grow in
their application and thus become harder to define as they become fundamental
to all automated IT management tools. The technology itself will also move
beyond just making sense of the data and informing humans to make better
decisions, and ultimately it will enable proactive automation with self-driving
optimization. However, the road to true proactivity and automation relies on
the effective implementation and integration of AIOps with deployed monitoring systems
and their ability to autonomously make decisions that lead to better business
outcomes. This will likely require a
‘crawl - walk - run' approach - the first phase of which I like to call
‘automation with governance'. Make the
recommendation, implement it and observe the outcome. Then confirm or refine the approach until
true autonomous automation can be unleashed.
VMblog: So, what can we expect next? What's
Virtual Instruments' vision for AI, ML and AIOps?
Gentry: Right now, we believe that implementing a
strategic approach to AIOps is going to require an ecosystem-centric approach,
which is why we are partnering with companies like AppDynamics and Cisco. At
the same time, we are working with enterprise customers to understand how they
can most effectively implement next-generation technologies like AI and ML in
their drive for infrastructure solutions like AIOps. Without that
understanding, it's going to be difficult for us to deliver and for businesses
to reap the full benefits of the technology.
We're already seeing that AI has become
more than just a buzzword, and 2019 is showing promise of being that year that
we finally cut through that hype and can start having real conversations about
how this technology is and can be used. Of course, we're interested in seeing
how AI can contribute to automating IT operations, but we're even more excited
to see how these technologies can help IT managers and their staff move beyond troubleshooting
and focus on more valuable initiatives that advance the business. From
improving efficiency and reducing infrastructure downtime, to being able to
deploy key resources beyond tactical needs and putting more energy toward
strategic tasks, we'll be able to see the true business impact of AI and
ML-enabled technologies across the entirety of the organization.
##
About John Gentry
John Gentry is an information technology
professional with more than two decades of experience in product marketing,
sales, and engineering. He's currently the Chief Technology Officer at Virtual
Instruments where he's responsible for keeping up with and understanding key IT
industry trends that affect product strategy and strategic alliances. John was
named one of eight global leaders inducted into the Information Age Data 50 and
has held various positions at mid-sized systems integrators, start-up
collocations, managed service providers, and established storage networking
companies.