Edward Wustenhoff, VP of Infrastructure and Platform Services, Juniper Networks

Today's Network Operations Are AI-Ops

I See IT Like IT Is OperationsAI & ML
Edward Wustenhoff Headshot
Today's Network Operations Are AI-Ops image

Self-driving networks are the future of IT.

In this final video in a series focusing on IT services (see parts one and two ), Juniper’s Edward Wustenhoff discusses AI-Ops, including how he defines it and what it means to him. Don’t miss this short, eye-opening video filled with real-world examples of how Juniper Marvis can dramatically improve network operations and help you successfully manage the availability, security, performance, and economics of your IT infrastructure.

Show more

You’ll learn

  • Why Edward believes we will have self-driving networks before we have self-driving cars

  • The keys to a successful AI-Ops deployment 

Who is this for?

Network Professionals Business Leaders


Edward Wustenhoff Headshot
Edward Wustenhoff
VP Infrastructure and Platform Services, Juniper Networks


0:00 [Music]

0:00 foreign

0:00 [Music]

0:05 I'm Edward today's topic is AI Ops and

0:10 what it means to me

0:12 you may remember that I'm a big fan of

0:14 Iron Man and there's a sequence where

0:15 Tony Stark is engrossed in deep thought

0:18 but certainly Jarvis interrupts and

0:21 refocuses him on something that really

0:23 needs his attention

0:25 that for me is a great example of the

0:28 essential function that artificial

0:30 intelligence should bring to it

0:33 it relates to iot and how that's so

0:36 different from the common internet of

0:38 people

0:39 the key observation is that devices

0:42 create so much more data and there's no

0:44 way people can keep up with it this is

0:47 where we need computers to help

0:49 I Define ai Ops as applying artificial

0:52 intelligence capabilities to manage the

0:55 availability security performance

0:57 economics and change of our

0:59 infrastructure

1:00 the ultimate goal is to allow for

1:02 self-driving networks or infrastructure

1:04 similar to how people talk about

1:06 self-driving cars but we are only at the

1:10 beginning of the capability and I'm

1:12 cautious about the claims being made to

1:15 that extent today

1:16 however when you can limit the number of

1:19 variables and the data the AI has to

1:21 process it is easier to build a

1:24 successful model and the key is to have

1:26 well-defined content which is the case

1:29 in network operations

1:31 this is why I believe we'll have a

1:33 self-driving Network before we have

1:35 self-driving cars

1:37 and today this is why Marvis is so

1:40 successful in helping us operate our

1:42 Network

1:42 let me give you a few examples it

1:45 improves our visibility into our

1:47 wireless network and provides proactive

1:49 resolutions for example we avoid not

1:52 knowing what we don't know it helps us

1:54 find Opportunities to improve

1:56 performance by showing where delays

1:58 happen for example just the other day it

2:00 showed us how a certain laptop's very

2:03 specific Wi-Fi device driver did not

2:06 play nice with our Network we now avoid

2:08 problems before they arise by

2:10 proactively creating a pick list of APs

2:13 that need attention it shows us

2:15 potential bad cables authentication

2:17 errors AP Hardware faults or other

2:19 network related issues

2:21 because of this visibility and this

2:24 predetermination of potential causes we

2:27 reduce the number of wireless tickets

2:28 over the last nine months by 42 percent

2:32 which is incredible when you think about

2:34 the fact that we already running a

2:36 Leading Edge Network

2:38 especially at scale this becomes

2:40 critical just think about the challenges

2:43 to resolve issues quickly in a world of

2:45 millions of iot devices thousands of

2:48 Home Offices or tens of thousands of

2:50 retail locations

2:52 and we recently upgraded 2 000 APS

2:55 across the world in a few days the

2:58 actual work of the tech was planning the

3:00 upgrade schedule to minimize the risk

3:02 and impact Marvis did all the work

3:05 the key to a successful AI Ops

3:07 deployment is being able to pierce

3:09 through the hype and test how much work

3:12 is left for humans after the computers

3:14 have done their job this doesn't mean

3:16 less humans but with Marvis keeping the

3:18 lights on it allows our team to add more

3:21 value and make more fundamental upgrades

3:23 to the network

3:24 Marvel's ability to analyze the data is

3:27 where we see the biggest benefit today

3:29 I'm not ready to eliminate the human

3:32 interpretation of the ai's observations

3:34 and rely solely on the suggested actions

3:37 for improvement yet however once that

3:41 confidence level is high enough I'll be

3:43 ready to hit the self-driving button

3:47 for our Network we clearly are on the

3:50 way and I expect more exciting progress

3:52 soon so if you would like to add ask or

3:55 a book a live demo let's talk about it

3:59 [Music]

Show more