All language subtitles for 17 - SAA EC2 Instances Launch Types Hands On English

So let's have a look at all the ways

we can launch our EC2 instances.

And the first one is using a spot request.

On the right-hand side I click on Spot Requests.

And we can have a look at first the pricing history

for our EC2 instances.

So if we look for example, at a C3

or a C4 large, for example, you're running Linux Unix,

and I'm going to look at three months range.

We can see how the price has evolved over time.

Based on April, May, June,

and we have the on demand price as a black bar right here.

And then we have based on the different AZ, the price.

And as we can see,

well, the price has been quite low and quite stable,

and based on the different range you're looking at,

you're looking at savings of around 69 to 70%

for this type of instance.

So huge savings if you're using a spot instance

of type C for large.

So how do we request a spot instance?

Well, we click on Request Spot Instances.

So I am in my spot fleet request screen

and there's a lot of parameters that I can see

but I wanna go quickly over them just so you get an idea

of how spot fleet request works.

So you can either use a launch templates

or manually configure launch parameters

and you can specify Amazon Linux 2,

the key pair you wanna use,

as well as some additional launch parameters

you would specify if you were to create an EC2 instance.

So this defines how your EC2 instances are created

and this makes sense.

Next, we have request details.

So we have more settings in here.

We can say, what's the max price you wanna have?

When is the request valid from and valid until?

Do we want to terminate the instances

when the request expires or not?

And are we linking these EC2 instances

to one or more classic load balancer or a target group

to have it, for example, for an application balancer?

So we'll use the defaults for this.

Then for target capacity, how many instances

do we wanna have as part of our spot fleets?

So we can say for example, 10 instances, okay.

We could also specify it in terms of vCPUs

if your unit of compute was more important

or megabytes of memory if RAM was more of your target.

So we can say, for example, we want 10 VCPUs.

We want to maintain the target capacity.

And then what happens, well, in case of interruption,

please terminate my instances

but you can also stop them or hibernate them.

Okay, and you can also rebalance the capacity if need be.

So this is just some tweaking around the target capacity.

Then for networking settings.

So where do you want a launch time,

in specific AZs or not,

in a specific VPC or not.

Then for the instance type,

do we want to have some restrictions?

So you can manually select the EC2 instances

that are matching some specific criteria

for your spot fleet request,

for example C3 large, C4 large, and so on,

or you could specify just instances attributes.

And this could say, hey, how many VCPS you want min and max,

how much memory you want men and max,

and then other attributes if you wanted to.

And then based on what you specify right here,

it's going to give you the matching instant types.

And we'll say the more restrictive

you are around these parameters,

the less instant type matches you're going to have.

But the idea is that you want to just launch whatever

AWS has in store for you so that you maximize the savings

while having some compliance

with the restrictions you set before.

Then the allocation strategy.

So do you want to optimize for capacity, so to make sure

that you have the right capacity all the time

or do you wanna have the lowest price

and you're going to get the strongest savings based to that.

And then if you had manually selected instance types here,

for example, one, two, three, four,

then you could select this option right here

to maintain a diverse pool of instances

across all the instant types you selected right here.

So once you've done it, you get some summary

of your fleets request.

And whether it's a strong fleet,

how many instances match and so on and which AZ.

And then it gives you the estimated hourly price

for this fleet, so 0.156 per hour at target capacity

which is going to give me a 73% savings

compared to on demand.

So no need to know exactly how these work, okay,

but just you can see all the parameters you can have

to do a spot fleet request.

If you want to launch a spot instance directly,

just go to Instances and then Launch an Instance.

And in there you can scroll down

and you will find

under the advanced details

the option to request a spot instance.

So if you do so, then you are requesting spot instances.

And by default, the request of the spot price

is capped at the on demand price,

which is quite good but you can customize this.

So you can, for example, enter your own maximum price

if you wanted to per instance, per hour.

Then you can choose the type of request you want.

So by default, it is a one-time request.

That means that whenever you request a spot instance,

in case it gets reclaimed,

it will automatically be terminated.

This is the only correct interruption behavior

or you can choose a persistent type of request.

In that case, then the spot instance

will keep on coming back

when the price requirements are set.

In that case, you have to set a validity of your request.

You say, okay, this request is persistent

until a date or just no expiry,

as well as what happens in case the price

goes over the maximum price you have.

In that case, you have an interruption behavior

of either hibernate or stop.

And finally, you see here block duration.

This is going away on December 31st, 2022.

So I'm not describing it to you right now.

This is a outdated future that you may not even see

on your account right now.

Next let's have a look at all the other options

available to us.

So we have reserved instances

and reserved instances allows us to purchase

a specific reserve instance of a specific type.

For example, we can look for a C5 large

and we can do a search of all the offerings.

And as we can see,

we have different offerings available to us.

For example, we have a 12 months or a 13 months term

and it could be a standard or a convertible type

of instance, and this is all upfront, but we scroll down

we can get all upfront, we can get no upfront,

we can get partial upfront.

We say how many you want, for example we want two of these.

So I'll just have the number two in here.

And then on the right hand side, you can just click

on Add to Cart to add this reserve instance to your cart.

If you click on View Cart, you can have a look

at what type of reserve instance do you want

and then we can order all.

But if you click on that, you will pay a lot of money

so I would suggest not to do this.

As we can see, reserve instances I think will soon go away

because now the AWS Savings Plan

maybe something that's easier to use,

and so this is why it's recommended to use the savings plan.

So the savings plans is available here on the left hand side

which will allow you to dedicate

a specific amount of dollar per hour you spend

on a one to three-year term and gives you flexibility

of how you want to launch your instance.

It gives you flexibility in terms of instance type,

AZs and so on.

So we'll not go over it but it gives you the option.

Next we have dedicated hosts.

So dedicated host is for you to launch a host

that you can access to a lower level kind of hardware

which gives you better licensing pricing.

Which is why if you click on this,

it will direct you to a license manager

which is a way to think in terms of licenses

and not think in terms of hosts.

And in which case it's going to simplify

the management of your dedicated host.

But if you wanted to launch a dedicated host directly

from this UI, you click on Allocate Dedicated Host

then you would name it.

You would specify the instance family, for example C5,

you would specify an AZ.

So on this dedicated host of type C5

we can launch multiple instance types

and then a bunch of settings as you go.

And then you click on Allocate

and you would have a dedicated host.

But again, I will not do this because it will cost you

a lot of money.

Finally for capacity reservations,

this is when you want to ensure that capacity is available

within your EC2 instance launch.

So for example, we're saying, hey, I want to make sure

that I have enough M5 2X large available to me.

So I want them to be in EU central one.

And I want them to have four EC2 instance

of type M5 2X large available to me no matter what, okay?

And then you can say when the reservation ends manually

on a specific time and then this will guarantee

that the instances will be available to you

but you will have to pay for this reservation

regardless whether or not you are launching

your instances in it, okay?

So that's it for this lecture.

We've seen all the ways you launch your EC2 instance types.

I hope you liked it, and I will see you in the next lecture.