Google Associate Cloud Engineer – Pricing Calculator

1. Step 01 – Getting Started with Google Cloud Pricing Calculator – GCE VMs

Welcome back. In this step, let’s look at pricing calculator. Pricing calculator is one of the best ways in Google Cloud to estimate the cost of a solution. Estimating the cost of a Google Cloud solution is not easy. You need to take a number of factors into account. So what is the easiest way to estimate the cost of your GCP solution? It’s to use the Google Cloud pricing Calculator. It provides you with estimate for a number of services compute Engine Google Kubernetes Engine Cloud Run App Engine Cloud Storage for most of the services in Google Cloud, you can go into pricing calculator and try and estimate what would be the cost of your solution. Remember, these are just estimates and these are not binding on Google Cloud platform.

So let’s get started. If you do a Google search for Google Cloud pricing calculator, it would be one of the first results. So I’ll pick the link from here. Google Cloud Platform Pricing Calculator the pricing calculator will take a little while to load. If you have problems in loading up the pricing calculator, you can also try it with a different browser. Let’s look at the different things that you can estimate cost for using Google Cloud pricing calculator. You can see that you can do pricing calculation for Compute Engine google Kubernetes Engine Cloud Run App Engine Cloud Storage you’ll be able to see the databases, the networking solutions, so almost every solution that you can think of.

So the big data solutions like Dataproc Database Cloud, SQL, Pub Sub Cloud DNS for almost every service in Google Cloud, you can estimate the cost of your solution in here. So let’s go in and start with Compute Engine. Compute Engine is the most basic service which is provided by Google Compute Engine. It is the IaaS infrastructure as a service solution. You can create virtual machines using Compute Engine if you want. You can even search for the product in here. So if I’m actually interested in GKE, I can type in GKE and pick up GKE. For now I would want Compute engine. It’s already selected. Or I can search the compute engine in here. Either way we can go down and we can say how many Compute instances that I would want to create.

So I would want to create ten Compute instances. I mean, this is just a description. What are these instances where you don’t need to provide the description? It’s not mandatory. The next thing you need to choose is the operating system. Do you want a free one or are you going for a paid operating system? For now, let’s stay with free operating system. The next important thing is the class of the machine, whether it’s regular or preemptible. So are you creating preemptible machines or regular machines? Preemptible machines are cheap, but they are not always available and Google Cloud can terminate them at any point in time. I’ll choose regular for now and the next thing you choose is the machine type family.

Are you going for a general purpose machine type family or computer optimized or memory optimized or accelerator optimized? Let’s stick with general purpose for enough and let’s choose a series. Which series you’d want to go for? You can choose a machine type. You can choose where you’d want to create your virtual machines in. This is the region in which you can create the virtual machines in. The price will vary a little bit based on the region. Also the next thing is are you making use of ephemeral public IPS or static public IPS? The next thing is are you going for committed usage? I can say I would go for three years commitment or one year commitment based on it. Your cost would vary. For now I’ll leave it at none. And next you can specify how many hours per day your server would be running.

So 24 hours per day and then you can say how many days per week your server will be running. If you are not running them on weekends, then you can put five in here. If you are running it on all the days, you can put seven in here. One of the important things to remember is you will need to pay for networking as well. There are specific situations when there would be egress costs. Let’s not worry about them for enough. Next, you can configure sole tenant nodes. Do you want to make use of sole tenant nodes? Nodes which are dedicated for you sol tenant nodes are obviously expensive. So if you want to actually use sole tenant nodes you can actually configure the sole tenant nodes that you would want to make use of.

I want to use five sole tenant nodes and you can configure everything related to those sole tenant nodes as well. For now, I’ll not worry about sole tenant nodes. I’ll go to persistent disk. Persistent disk is the place where you can configure the total volume of persistent disks that you don’t attach with your virtual machines and you can also choose the region. So I’ll take the default region and let’s say I’m going to use Zonal standard persistent disks and I would say I would want 2000 GB of that storage over here. We are not configuring how many we would want, we are configuring how much storage. If I’m attaching ten persistent disks with my virtual machines.

What would be the total storage that would be present on their persistent discs? You pay for persistent disks based on the storage and at the end you can also configure cloud TPUs. Cloud TPC are very very useful for machine learning workloads. So if you’d want to use any cloud TPUs, you can configure them in here. For now, let’s not really worry about them for now. What I would do is I would say add to estimate for instances. So instances are now added to the estimate and I’ll scroll down and I’ll say add to estimate for persistent disks as well. Let’s do an add to estimate in here as well, and these are added to your estimate. The estimated cost for Compute Engine is about $500 per one month.

This is to run ten Compute Engines and you can see that the estimated cost for your persistent disks is about 569 USD per month. Now, if you don’t, you can go in and edit and see what would be the difference if you want to make use of some other option. So I’ll go in here and I’ll say I want to make use of, let’s say, the preemptible machine class. I can see what would be the difference. I can say add to estimate and you can see that now my costs have dropped down drastically. I only need to pay $146 per month. So you can actually try playing with it. You can try different options, different machine types, and see what would be the cost. Now, in addition to Compute Engine and Persistent Disk, if you want to add any other things.

Let’s say you’d want Estimated Database or a Kubernetes Cluster, you can add all of them to your estimate. And then at the end you have a total estimate for your solution. In this step, we looked at how you can use Google Cloud Pricing Calculator to estimate the cost of your solution. In addition to enabling you to estimate a solution, I find that Google Cloud Pricing Calculator is a great learning tool. If you come over here and look at Compute Engine, you can get an overview of what are the very, very important choices that you would make when you create a Compute instance. You can hover over this. If you have a question, you can actually hover over this and get more details about that. So by doing this, you can quickly review things at the time of an exam.

So just before going to the exam, one of the most important things I would recommend you to do is to come over here, come to the Pricing Calculator, and look at the different factors and make sure that you’re understand them. If you don’t understand them, you can actually click the question mark in here and look at that in specific detail. What we’ll do in the subsequent steps is look at a few more services in Google Cloud Pricing Calculator. But remember that just before going to an exam, just spend some time with the Google Cloud Pricing Calculator and look at the different options, because I find it as one of the quickest ways to reduce something just before an exam. I’m sure you’re having a wonderful time and I’ll.

2. Step 02 – Playing with Google Cloud Pricing Calculator – Compute Options

Come back in the step. Let’s look at a few more things that you can estimate using Google cloud pricing calculator. Let’s go to GKE standard. So how do you estimate the cost of your GKE solution? In GKE we have node pools. That’s what is the major cost as far as GKE is concerned. So you need to say, I have this many nodes in my node pool, I have ten nodes, and you can configure what are the Is for. I’ll take the defaults. You can configure if you’d want GPUs, you can configure if you would want committed usage and how long you’d be running these clusters. In addition to the node pool, you also have a master which is running. So you need to add that to the estimate as well.

So let’s first add the node pools to the estimate and then you need to account for the cluster cost. So cluster is where your masters run. So you can say, I have one zonal clusters, I would say. So you can see you can see that now you have the cost per GKE standard node pool and the cluster management fee also added in. In addition, if you are attaching any persistent disks with your node pools, you need to add that as well. So let’s say I would want 300 GB and I would add that to the estimate. So whenever we’re talking about Kubernetes clusters, the costs are one, the node pool cost, the cluster management fee, and also the persistent storage. Because we already had persistent storage earlier, it was added into that.

The 2000 from earlier step is now bumped up to 2300. And that’s what you’d see in here. The other solution that we talked about earlier is Cloud Run. So Cloud run, you can also estimate the pricing for it in here. So you just configure over here. You need to configure how many CPUs are allocated, how much memory is allocated, how many concurrent requests do you think one container instance can handle. So if you have a microservice that is deployed in cloud run at a particular point in time, how many requests, how many concurrent requests can you handle per container instance? That’s an important thing that you need to configure in here. And then you can configure how long do you expect a specific request to the microservice would take and what is the number of requests that you expect based on which cloud run would calculate the price of your solution? Another interesting thing is app engine.

We looked at App Engine earlier and you can see that there is different pricing for Standard and flexible. As we discussed earlier, App Engine standard does not provide you with a lot of choices. You cannot configure the hardware of the instance types that you would want to make use of. And that’s the reason why pricing for it is also very, very simple. All that you need to say is I would want to run it in this location. And this is the type of instance. So I want to use F one or B one, b two. What is the type, what is the machine type that I would want to make use of? So that’s the second criteria. That is the second thing that you would choose and the third thing that you’d to choose is very very simple.

How many instances do you want to run per hour? So I’ll say ten instances. That’s it. Calculating the cost of standard environment is very simple. However, calculating the cost of flexible environment instances is a little bit more complex. If you are using virtual machines to run your flexible environment instances, you need to configure how many vCPUs are attached to your group of instances, how much memory is attached, how much persistent disks are attached. Now, there are a few additional costs that are associated with App Engine if you are making use of a few specific services. So if you have traffic going out to Internet from your App Engine instances, then you’d need to pay for it.

You need to estimate how much traffic you expect to the internet from the App Engine. Remember, this is egress traffic, this is outgoing traffic. We are not talking about incoming traffic over here. If you are sending any requests to Internet from App Engine, that’s what we are paying for. In addition to that, if you are making use of any services like cloud storage, mem cached or logs or task queue, you need to add those things to the estimate as well. In the step we quickly looked at using the Google Cloud pricing calculator for the other compute solutions GKE, Cloud Run and App Engine.

3. Step 03 – Playing with Google Cloud Pricing Calculator – Databases

Come back. Instead, let’s start playing with pricing calculator and look at the pricing for different databases. Let’s start with BigQuery. BigQuery, as we discussed earlier, is a data warehouse. You’ll store a lot of data, you will query it. And as we discussed earlier, BigQuery also supports streaming workloads. So let’s see how the cost for BigQuery are determined. You can either go for on demand or flat rate. In flat rate, you’d need to book a few slots. So you can pre book slots and that would give you a discount. This is very, very similar to committed use with virtual machines. But if you don’t expect a huge amount of volume or work, you can go to OnDemand.

You can say, this is the specific table name, and I can configure how much storage I would want to make use of in BigQuery. How much data will be stored? Let’s say it’s 2000 GB. You need to specify how much streaming data comes in. You can use BigQuery to store streaming data as well. How much volume of streaming data is going to come in? I’m saying 200 MB. And the next thing which is important is the query pricing. How much data will be scanned by your queries? Remember, this is not the data which is written by which is written back from a query. This is the amount of data which is scanned to return the result. How much data needs to be scanned? That is what is present in here. Let’s say I would process 2000 GB of data. Let’s say I would process two terabytes of data.

Then I can add two in here, and at the end I can say add to estimate and this would give me the estimate for my BigQuery solution. So let’s scroll little down and see BigQuery solution costs. You can see that this costs about 600 per month. Now, let’s look at pricing for other databases. Let’s go to data store next. As we discussed earlier, Datastore is a NoSQL database and the pricing for Data Store is very, very simple. How much data are you going to store? 100 GB. This is the Reads per month, Writes per month and Deletes per month. Just put in those numbers and you should be able to calculate the pricing for Data Store. Five.Store also is similar. However, over here it’s the document reads, writes and deletes per day estimate that you are going to give along with the amount of total stored data.

The next one is Dataproc. Dataproc we talked about is the Managed, Hadoop and Spark service. If you have Spark or Hadoop based applications that you don’t want to run in Google Cloud, you can use Dataproc. And we talked about the fact that Dataproc uses clusters and inside the clusters you have virtual machines. So when you are using Dataproc, you need to first define your cluster. What is the type of machines that are present in your cluster? How many you would want and the calculation would be based on that cloud SQL. How do you determine the cost? Cloud SQL also runs on compute engines. And that’s the reason why you need to say how many instances you would want and you need to configure how many virtual CPUs.

How much memory you are making use of, how much storage you are making use of, are you registering for a committed use, and the running time of your database. In addition to that, if you have automated backups, then you need to pay for the automated backups as well. Let’s go to Cloud BigTable. BigTable, as we discussed earlier, is another no SQL databases which is recommended when you have a lot of streaming data which is coming in to the volumes of Petabytes and in big tables. What we do, we create clusters so you can give a name to the cluster and you can say how many nodes will be in the cluster and you can give how much SSD and Http storage that you’d be making use of. You can expect the big table nodes to deliver up to 10,000 queries per second and ten MVPs of throughput.

The next database that we’ll be looking at is Cloud Spanner. We talked about the fact that Cloud Spanner is a relational database which is recommended for global workloads. If you have large relational databases and you want to distribute them across the world, cloud Spanner is a great solution and the price is based on whether you’d want a regional solution or a multiregional solution and you can choose how many nodes you would want and how much storage you will be making use of in this. If we looked at pricing for some of the important databases in Google Cloud, I’m sure you’re having a wonderful time and I’ll see you in the next next step.

4. Step 04 – Playing with Google Cloud Pricing Calculator – Others

Welcome back. In this last step on Google Cloud Pricing calculator, let’s look at a few more services that you can calculate the price for using the pricing calculator. Let’s start with cloud storage. Cloud storage is object storage. In cloud storage we have buckets and we can upload objects to the buckets and retrieve objects from the bucket, and therefore you pay by how much storage you are making use of, how many GB’s of data you are putting in cloud storage. And there are two types of operations class A operations and Class B operations. If you click this link and go over to the pricing page, you’d be able to see a list of Class A operations and Class B operations. So if you scroll fully down, you can see a list of Class A operations and Class B operations.

You can see that insert update, changing the Im policy, copying objects, listing objects. All of these fall under the Class A operations. If you are retrieving data, most of them fall under the Class B operations. So you need to give an estimate of how many operations that you’d perform in Class A and Class B and based on which your solution cost will be determined. The next important thing is networking egress. You need to pay for the internet traffic. So from your source region, which region are your resources in? How much Internet egress do you expect to different parts of the world? You need to configure that in addition to the internet traffic, you will also need to pay for VM to VM egress.

How much communication is happening between your VMs. Important thing to remember within the same zone, if you are using internal IP addresses, then VM to VM egress is zero cost. However, if you are communicating to another zone in the same region, or if you are communicating with a VM in another region, then you need to pay for it. So you can choose how much egress is going from your virtual machine in this specific region to different zones or other regions. In addition, networking costs also include interconnect egress. If you have a hybrid cloud setup where you are using cloud interconnect to connect your data center to the cloud, then you need to pay for that as well.

The next thing that we look at is data flow. The data flow, as we discussed earlier, is used both for streaming and batch applications. If you want to migrate data from one database to another database, or if you have a lot of streaming data coming in and you want to store to a database, in those kind of situations, you can use dataflow. We talked about the fact that dataflow is serverless and auto scaling and therefore you’d actually pay based on the type of jobs that you are running. In last few steps, we are using Google Cloud Pricing Calculator to look at how prices are estimated for different things. What I would recommend you to do is to play around with it a little more, and I’ll see you on the next step.

• Category: Uncategorized