SKILupDay

Superpowered APM!

It helps orchestrate speed.

AI Ops uses artificial intelligence to enhance traditional IT and cloud native operations through data aggregation, automation, correlation, pattern recognition and orchestration.

Application of machine learning practices to the field of IT Operations

Power of AI in Ops

It enhance predictive & adaptive decision making #AI

(edited)

MLOps is the capability to apply learning models to large datasets for IT applications. AIOps is the capability to generate decisions based on heuristic algorithms operating on large data sets and make use of MLOps to do that.

Great way to be able to be able to forecast and identify issues relating to performance and bottlenecks in your applications #SKILupDays #AIOps

in simple words, Artificial Intelligence for the Information Technology Operations. #AIOps #MLOps #DevOps #SKILupDay

AIOps: For me AI-based Algorithms for ITOps use cases with goal to reduce MTTR, MTTD by iteratively increasing the quality of resolutions

I would define AIOps is using of the large amount of data that is generated and collected in Operations to help making the systems more stable and reliable.

@mhexcalibur precise

@DEVOPSINST, #SKILupDays #AIOps: AIOps as Gartner defines stands for "Artificial Intelligence for ITOperations". It also refers to multi-layered technology platforms that automate and enhance IT operations through analytics and machine learning (ML).

AIOps mean replacing your developers with agents who can code directly for you, don't take vacations, and only rarely ask for raises

@mhexcalibur As long as AI generates some type of decision, does it matter what basis is used? I suppose all decisions are algorithmic but is there a need to check the algorithm or just inputs/outputs to AI-box?

data versioning and model repo will be some new features for DevOps

In short it help you to become more proactive, predictable in your business. To improve the pace and maintain the pace of Ops its very important to monitor, predicate, detect and respond to failures in efficient way and Al, ML can helps and makes it more faster

Depends on the following
1. Do you look for predictive & adaptive enhancement in flow

It is another advantage factor for optimising performance, being able to use your infrastructure better and optimise your applications to be able to have that competitive advantage #SKILupDays #AIOps

2. Do we have the right data to learn from (feedback)

3. If experimentation & innovation takes the centre stage

Reduction (or elimination) of MTTR gives time back to teams to innovate

We are working with highly complex system that generate large amounts of data. Our compute capabilities have increased enough to run more complex algorithms against that data to gain insights and make decisions.

AIOps and MLOps have the potential to revolutionize DevOps. The insights and real-time responses derived from being able to ingest and intelligently analyze large data sets enable IT and DevOps applications not possible with limited capacity human processors.

@anuragthoughts Excellent!

AIOps enables to enhance decision making and sometimes even automate them. MLOps creates rigor around the creation and automation of the models we use every day. Both have the potential to advance what we consider DevOps today.

To connect the entire journey of enterprises starting from concept to realization leveraging cognitive analytics thereby driving business outcome

AI and ML are already in your day to day functions. New tech, new storage space, and faster computers means the ability to handle new questions faster. Quantum computing means if you aren't already ahead of your ML, may be behind.

thanks to Covid, we're now at an accelerated digital transformation era. So, these #AIOps and #MLOps are not a 'nice to have' any more, but a 'must have'.

People first, then processes. Advanced #AIOps and #MLOps are a nice to have, some aspects are incorporated without ever having to think about AI. Think about Siri, and google search...

@BealHelen @DEVOPSINST And reducing MTTR is critical to creating business value, faster recovery means better time to market.

- Few reasons are; how do we monitor the health of the IT landscape that is dynamic with the blind spots?, how can we predict failure and their business impacts?, visibility across cloud environments (on-prem, public and private included).

one of the problems with AI libraries is that they do presuppose that individuals can code. I expect that most of the AI will, in fact, be embedded.

Hi All...

Hello!

(edited)

Hola

Hello Everyone

Hello!

looking forward to it #AI

Happy to join in the discussion

With the right data, you can put AI anywhere within the process, training inputs = valuable outputs.

AIOps can be used by autonomous multifunctional teams to not only accelerate incident management but also to measure value realization via customer experience

Operations is a great place to start as it has the data and a problem to solve.

AI can be applied to continuous testing. Select tests and craft test suites dynamically to match risk or code changes.

@DEVOPSINST AIOPs can be applied to any part of SDLC process. Identifying the right use case and appropriate model implementation is important.

@BealHelen When we limit AI to vulnerabilities, we limit our perspective. While AI starts small, at some point need to think about using AI whether metrics aggregate across your processes.

AIOps best in facilitating case for Selfservice for request fulfilment tracks

Depends on business Questions

AI and ML is not as half as smart as people like to think, you need to think how to guide it and understand the results

My fav. myth more data better AI, ML. otherside AI can risk my data :)

-depends if you have a clear business question or not :-)

AI and ML Ops are not fire and forget. Models need training.

AI, as we define it today, is not "a computer that thinks like a human" it is really the application of deep learning on large datasets.

Mythbusting: Just adding AI/ML to your process will not solve all your business problems.

All data is useful for ML, ML make systems exactly think like humans, AI will be bias-free and AI deployment urge a huge cost for desired outcomes are in top of my mind.

@lynxman Somedays, it feels like neither am I or the people I work with, somedays we are all twice as smart as the AI

AI is not yet true AI. See the reply from @lynxman earlier

Myth: AI and ML process too slowly to be useful during rapid, incremental software delivery cycles. This is a myth because the algorithms and knowledge generated by AI and ML can be called up and used for DevOps applications on-demand.

@BealHelen And the reply from @pgmaddison too

@bajpaigarima1 @DEVOPSINST I'd like to have the AI that can work with unclear business processes, the clear ones I can handle.

#AIOps #MLOps are the buzzwords and there isn't anything in near future that organizations should worry about. Whereas the fact is that AIOPS is very likely to become the foundational building block/system for sustaining, managing, and transforming the infras #DevOps

Myth: the data our organization has will be sufficient to train my ML/AI Agent

Myth: Data cleaning and preparation is too expensive, too complicated and time consuming.