Getting Started with Cloudwatch
A free video tutorial from Ryan Kroonenburg
4.5 instructor rating • 10 courses • 710,597 students
Learn more from the full courseAWS Certified SysOps Administrator - Associate 2020
Want to pass the AWS Certified SysOps Administrator Exam? Want to become Amazon Web Services Certified? Do this course!
16:11:50 of on-demand video • Updated February 2020
- Pass the AWS Certified SysOps Administrator 2020 - Associate Exam
- Become a certified AWS SysOps Administrator
- Become Intimately Familiar With The AWS Platform
English [Auto] OK hello cloggers and welcome to this lecture. So this lecture is on cloud watch and cloud watch is a fundamental part of this ESOPs administrator associate exam. Basically it's one of the most important services that you need to understand inside out when going to sit your exam. So we are in the monitoring and reporting section of the course so we're going to look at cloud watch and how we can do monitoring and reporting. So let's start with what cloud watch is basically it's a monitoring service to monitor your AWOS resources as well as the applications that you run on yes and cloud watch can monitor things like compute kimonos and things like your auto scaling groups or elastic load balancers route 53 health checks and reduced storage and content delivery so abs volume storage gateway cloud frumps cetra can also monitor things like your database and analytics so these are things like dynamite D-B or less to cache nodes ADX instances your lastic map reduce job flows redshift etc. and it can also monitor other things such as this and estop exists and eschewed artworks cloud watch Lokes estimated charges on your A-type U.S. bills et cetera. So the most fundamental important thing to understand going into your exam is what it monitors by default when it comes to Acey too. And then what you need custom metrics for. So by default it monitors these things called Host Lebel metrics and they consist of basically four different things three different things from a physical point of view and then a status check. So the three different things cepa you network and disk so components that things like CPQ utilization or network throughput or disk IO for example. And then it can also monitor the actual health of both the PC to host as well as the PC to instance. These are known as status checks. What it can't monitor is things. So if you think about it holistically can see the disk as a whole as a physical piece of hardware. What it can't see is the individual. Let's say we go down to the individual virtual disk because see how much storage space is left or is still available. It can't see Ram utilization can't see things like that. What it can see is the C.P utilization. They can see the things like the network utilization and the overall disk activity so the overall disk throughput. But that's where it stops. And then it can also do things like your status checks. So anything else that falls out of C-p network disk will status check is a custom metric. So things like Ram utilization for example how much memory is your easy to instance currently utilizing all of which has provision to it and so that is what we call a custom metric. And we will have a lab coming up which will show you how to monitor this. We're going to use a postscript to go in and configure it on houri see to it. Since then we're going to start pushing this these metrics to cloud watch. So just remember going into your exam if it's not CPQ network or disk or the actual status of the two hardware sorry the easy to host as well as the PC to instance then it's going to be a custom metric and you will be tested on this through out the exam. Also it's really important to understand how long cloud watch metrics are stored. You can retrieve data using the Get metric statistic API or by using third party tools offered by AWOS partners and you can store load data in cloud watch logs for as long as you want. By default cloud watch logs will store you'll load data indefinitely so you can change the attention of the group at any time you could just say ok for my EO Bays I only want data for the next. For the last two weeks but for my two instances I want to go all the way back to six months so you can change it but by default it is logged indefinitely. And you can also retrieve data from any terminated E.C to or lastic load balance instance off to its terminations so do bear that in mind as well. Terms of metric granularity it really depends on the AWOS service. Many metrics for many default services. One minute but it could also be three minutes or five minutes depending on the service. Now you will get a you get basically normal monitoring and detailed monitoring normal monitoring is every five minutes detailed monitoring is every one minute and custom metrics so things like Ram utilization which we'll see coming up the minimum granularity that you can have is one minute. Now one important thing to note here is if you have just your default cloud watch settings on flight say AC 2 you're only ever going to get five minute data points. That's even if you do have a script that's pushing data to cloud watch every one minute cloud watch will actually aggregate that over the five minutes and again we're going to see that in the lab. Coming up terms of alarms you can create an alarm to monitor any Amazon Cloud watch metrics in your account and this can include easy to C-p utilization elastic load balance and latency or even the charges on your AWOS bill and you can set the appropriate thresholds in which to trigger the alarms and also what kind of action take in if an alarm state is reached and we're going to cover this off in a subsequent lecture. But you could just tell it to send you an S and S notification saying oh look up the owner of a $10 Oh my NWS account for this month. Or you could even have it trigger lambda function which goes through and just starts deleting all your infrastructure it's entirely up to you. But Cloud watch alarms can basically trigger an action to be taken which you would go in and configure. And yeah it's really really powerful stuff. So onto my exam tips just remember what cloud watches it's a monitoring service to monitor your AWOS resources as well as applications that you run on AWOS your host liberal metrics oyes consist of you network disk and your status check if it's asking you about a metric outside of this such as Ram utilization or even the amount of storage space left on a virtual hard disk. You're going to need custom metrics or ram utilization is a custom metric. I remember that full custom metrics the minimum granularity is one minute or so moving on terminated instances you can retrieve data from any terminated AC two or elastic load balancing instance after its termination and cloud watch Loeb's by default a stored indefinitely. And then finally just remember with your metric granularity So you have one minute for detailed monitoring and five minutes for standard monitoring. You may get some scenario based questions just asking you for two minute monitoring. And how would you achieve that. Well you're going to need to use detailed monitoring rather than standard monitoring. And one thing I didn't really cover off in this lecture. But one thing you should definitely be aware of is that cloud watch can be used on premise as well. It's not restricted to just AWOS resources. You can definitely use this inside your or your own data center and you can have the metrics being reported back to your cloud watch dashboard you know inside AWOS from your own data center and the way you do this is you need to download and install the SS agent and we will Kubla SS am in a lot more detail later on in the course. Then once that's installed you need to download and install the cloud watch agent. So if you do see any scenarios where it's talking about cloud watch being on premise that is definitely valid. So that is it for this election guys in the next election we're going to go ahead and start getting our hands dirty with the cloud. So if you've got the time please join me in the next lecture. Think.