[Music] hello my name is jason smith and i'm the outbound product manager for itom visibility at servicenow today we're going to show you how we use machine learning to make service maps getting started with machine learning driven mapping is very simple with servicenow the first step is to set up discovery schedules so that you can find the configuration items in your public cloud and on-premises environments once we get the data into the cmdb machine learning can start to take action so there's two major things that are going to happen one we will use the data to identify applications running an environment and we use a technique called application fingerprinting and then we will not only understand the connections between the applications but also the importance of the connections between applications and that will assist you greatly in making your service maps once we get the applications and services in place then we will make sure that we are conforming with the common service data model and we'll use that to power multiple workflows to set up the discovery schedules you can navigate to discovery home and then schedules and in this case i've got several different schedules already set here some of them are ip based and some of them are public cloud-based so if i want to edit one of these just click on that and i can choose a mid server the account information is already set up i could be using aws iem roles instead of this account information in this fashion but what i want to do is discover all of these data centers and all of the configuration items in those data centers and then importantly i would like to discover the virtual machines so what's happening is we're doing a public cloud discovery getting all the ip addresses and then we're doing a deep dive discovery on all of the servers running in that virtual environment as part of the server discovery process we're picking up processes running on machine and all of the tcp connections that the processors are making that information is sent over to machine learning for analysis and the result is that many of the applications running in the environment have now been fingerprinted by machine learning in addition to doing application fingerprinting the entry points to the application services are also discovered so in this case we've got 25 different entry points i'll open up one of these and i can see that this was an xa proxy running at this ip address load balancer so most likely this load balancer is in front of an application server so let's click on discover start that process and then i will go to the application services this is the one we just created in that fashion let's take a look at the map it says it's in progress okay so the initial pass is done so what we have here is the aj proxy and then we've got two homegrown applications that were originally identified by application fingerprinting and i can open this and look at the connection suggestions if i want i can add those that will add them to the map and then if there's more kinetics and suggestions i could look here in this case there's two more let's go ahead and add those okay so now we've got a map of a micro service application that's talking to different database servers on different machines you could change the view here and look at the host who you don't want it's two different servers that play here okay very easy to make these maps with the machine learning assisted help so i'm now satisfied with the application service map i would like to bring it into the common service data model so go into the navigator find the common service data model application scroll down to manage technical services application service and then we can see this is the service that we just met open that one up let's give it a better name call it learning and i would like to connect this application service to a business service offering and then it's a matter of just following the wizard updating that okay so let's take a look at the dependency view and we have a top level business service called shenandoah and under shenandoah i have a technical service offering and a business service offering this is the service offering for education and then here is our application service called learning and i can drill into that and take a look at the eventual related services like linux or this application service only map here so when i click here that's all of the components that are being used by that service and i can drill into the application service itself so very easy to get the services map with the help of machine learning we're identifying the applications through application fingerprinting we're identifying the important connections between the applications to make up the application service and then getting ultimately a good performance with the common service data model thank you