In this course, you'll learn what Elastic Compute Service (ECS) is, how to use it, and what makes Alibaba Cloud's new 6th generation instances faster and more powerful than their predecessors. You'll also learn how to choose the right type of instance for your workload, and how to save money by understanding the ECS pricing model. Finally, you'll learn how to launch your own ECS instance from Alibaba Cloud's web console.
Learning Objectives
- Understand the fundamentals of Alibaba ECS
- Learn the advantages of the sixth-generation ECS hardware
- Learn the instance types and families in ECS
- Learn how to choose the right instance type for your workload
- Understand ECS's different pricing models
- Learn how to purchase an ECS instance from the Alibaba Cloud console
Intended Audience
- Solution architects
- System operators
- Developers
- Anyone who wants to learn about Alibaba ECS
Prerequisites
To get the most out of this course, you should have a basic understanding of the Alibaba Cloud platform.
We now come to the most important topic of all, Understanding ECS Pricing. How do you get the best value for your money? So we'll start by taking an overview of some of the pricing modes on Alibaba Cloud, starting with reserved instances. So reserved instances or RIs, offer a discount based on a usage commitment. So you commit to use Alibaba Cloud resources for a certain amount of time, and then we provide you with a discount.
So the RI or reserved instance is a hybrid between pure pay-as-you-go and pure subscription. So originally Alibaba Cloud only offered two models. Subscription is where you pay up front for a month or several months for a year. Pay-as-you-go is where we charge you hourly. And then give you a bill at the end of the month, based on your hourly usage. RI is sort of a hybrid between those two. The issue is, if you're using pure pay-as-you-go, you get no discount for using it long-term. But if you're using subscription, you get a discount, but you're not able to flexibly delete or move your instances to another zone.
The subscription is tied to the instance and you can't delete the instance until the subscription expires. So how can you get the benefits of a subscription discount with the flexibility of pay-as-you-go? That's what RI does. Essentially RI acts as a coupon. When it pays you go instance is launched payment will be deducted directly from a matching RI. So if there's a matching RI, then rather than being directly charged to your monthly bill, your pay-as-you-go charges will be applied against that RI and you'll be given a discounted rate.
Another big advantage is that the RI can apply to an entire region. So you can move your instance between zones, if you want to and you can split RIs up. So if you have an RI, that's good for an ECX G6 2X large, You can actually use that RI to launch two regular X larges. So you can actually launch two smaller instances using an ROI that was intended for a single bigger instance. There are three payment methods for RIs, no upfront, partial, upfront, and full upfront payment.
So the reason for the different upfront payment options is that we offer different discounts based on which choice you make. And you have three and five-year commitments. Five-year commitments aren't launched yet on our international website on alibabacloud.com, but they are coming soon. These long-term commitments will give you a deeper discount. Sometimes if it's for a three-year commitment, it can be up to 45% off. If it's a five-year commitment, it can be up to 60% off the normal monthly price.
So here's the pricing strategy and the level of discount that you receive for the one, three and five-year commitments. You can see that the discount scales both with the amount of time you're willing to commit and with how much upfront payment make. So if you make an all upfront payment with a five-year commitment to use those resources, you get a 62% discount off the regular price for the instance. So the all upfront, partial, upfront, no upfront, and one, three and five-year commitments give you a wide range. They give you a lot of flexibility in how you purchase your RI based on how long you think you're gonna use these Alibaba Cloud resources.
So what about network and storage? The RI gives you flexibility and long-term use discounts for compute, but it only covers ECS itself. It doesn't actually cover the cost for disk or network. So to achieve discounts for bulk usage of storage and network, we have two other concepts we need to introduce. One is the data transfer plan, and one is the SCU. So the SCU is a Storage Capacity Unit. This allows flexible discounts for longterm commitment to storage use. It's similar to an RI, but for storage. And the SCU plans range between one terabyte and 50 terabytes. And they can span either multiple zones in a region or multiple regions. And the best part is these apply not only to cloud disk, but also to Network Attached Storage, OSS and Snapshots. So again, just like with an RI, the longer your commitment, the higher the storage discount you'll receive.
Then there's the data transfer plan, this is for network traffic. This has monthly and annual subscription types ranging between 50 gigabytes of network traffic and 50 terabytes. And again, this applies to ECS public IP addresses, so ECS public network traffic, as well as our other network facing services like Load Balancer and NAT Gateway and Elastic IP. And again, these are designed to give you long-term committed usage discounts for storage and for networking. I should point out, by the way, that for SCUs and data transfer plans, the payment method is all upfront. So you do pay for these upfront. Whereas for RIs, for ECS, you can choose no upfront or partial upfront.
Then there are Preemptible Instances. These give strong discounts for batch workloads or occasional use scenarios. So what do I mean by an occasional use? While Preemptive Instances let you bid for unused Alibaba Cloud inventory, essentially they give you a big discount on the normal pay-as-you-go price. So the way this works is, Alibaba Cloud, as a cloud vendor, has to allocate large amounts of hardware in each of our data centers to make sure that we can meet demand whenever users want to purchase an instance. But what that inevitably means is that a large amount of this hardware is sitting idle, sitting unused, unused by us, unused by our tenants, our customers.
So what we want to do is increase utilization. And Preemptible Instances let us do that. Basically what we do is we let you submit a bid, it's a price per hour that you're willing to pay for this unused capacity. And if your bid price is higher than our asking price, then we launch your instance. Each instance is guaranteed to run for at least an hour, but after that, it can be reclaimed. We will give you a five minute warning before we reclaim the instance, but we can't take it back.
So these instances are really good for batch computing, stateless applications, non-production workloads. It's just a regular EC SVM, but at a very, very low pay-as-you-go hourly price. And you can change your bid if you want. So you can adjust your bid flexibly to match our asking price, if you want your instance to keep running. But essentially this is a way to get a big discount for batch or a temporary workloads. If you're curious about how the actual Preemptible Instance bidding process works, there's a flow chart here in the lower right-hand corner of the slide that shows the process from creating instance to termination.
Basically, if your bid, the amount you're willing to pay per hour, is higher than we're asking for and we have sufficient inventory then your instance will launch and will run for at least an hour. After an hour, every five minutes, we'll compare your bid to the asking price. And if your bid is no longer higher than the asking price, we may reclaim your instance. If your bid is greater than or equal to what we're asking per hour, then your incidents will continue to run either until you choose to shut it down or until we reclaim it due to your bid no longer being higher than our asking price.
Okay, so how do we choose the right model? You've seen Preemptible Instances, Reserved Instances and SCU and data transfer plans, how do we put these things together? Based on your workload, what should you choose? Well, here is a handy chart to help you out. So if you have concurrent workloads, like you can see at the top of the slide where you have several different workloads that are running at the same time and their usage is related, so when one workload's resource use increases the other workloads resource use also increases as you can see with the blue, orange and yellow bars on the chart.
Then you want to consider pay-as-you-go, or RI plus pay-as-you-go as your payment model. This way you can flexibly scale up and down. If you have a very steady long-term workload, you could use pay-as-you-go, or RI plus pay-as-you-go for some of your workload. But you should really consider either long-term full upfront RI or traditional subscription. For spiking or bursting workloads, the obvious choice is pay-as-you-go.
You could also consider Preemptible if some of these workloads are not time sensitive. And then if you have mixed usage, multiple workloads that are changing and are not directly related to one another, so some workloads need to burst now, other workloads don't, then you can consider either pay-as-you-go, or RI plus pay-as-you-go if your resources are all of the same instance family. Because the RI again, can be split flexibly for instances that are within the same family. So in general, just keep this in mind. If you have a steady workload subscription or long-term RI with upfront payment, if you have a changing or bursting workload, Preemptible plus pay-as-you-go.
Alibaba Cloud, founded in 2009, is a global leader in cloud computing and artificial intelligence, providing services to thousands of enterprises, developers, and governments organizations in more than 200 countries and regions. Committed to the success of its customers, Alibaba Cloud provides reliable and secure cloud computing and data processing capabilities as a part of its online solutions.