Hello, Siberians. Welcome to Lesson 3.9. Off Military Off Discuss titled Is that receiver one Microsoft Azure Architect Design.
Here are the landing objectives for this video.
We'll start out by giving an overview off Azure cosmos TB. Where exactly is this service?
Well, then proceeds to cover the different AP heist that as your cosmos TB supports
and we'll also cover the consistency levels that it supports
finally will give an overview off the resource. Iraqi, of course, Mostly be so that we understand the different technologists.
Let's get into this. So what is that? Your Cosmos D B
as your Cosmos D B is a globally distributed, massively scalable, low literacy multimodal database service.
But where exactly does that mean
it's a globally distributed database service? Because with a single click off a mouse, we can replicate at database tow any azure region around the world.
It's massively scalable. You see some of that in a minute when we get a design decisions. So, for example, the stoppage can be unlimited.
It has low latency because with azure cosmos d B. What you're actually painful is government ID performance to the team off less than 10 minutes. Second for night night. Peace ain't off right and read operations.
It's a multi model database service because we can select different data models on a P I access. For example, we can select the table FBI Cassandra Epi. I see quality high
cost must be be FBI from Mongo, D. B or the Kremlin. A p I.
There's some more information about the AP eyes that cosmos TB supports. It supports Mongo D B a p I,
which means that it's compatible with existing money. Would they be libraries? Drivers? Tooth. An application
on what? Our some misses. That existing Mongo D B databases can be migrated over to azure cosmos TB,
and you did not need to modify your applications cause you can use the same way a protocol to communicate with Cosmos TB.
It also supposed to table AP, and if you're familiar with, like the azure table storage, this is a similar thing in terms of the A. P. I. Itself is a key value database, but it's now against benefit off the enterprise capabilities off Cosmos TB,
the global replication capabilities, of course, most baby
on the low literacy capabilities. Of course, most EBI
It's a plastic gremlin. AP High, which is based on a patch it in cup of specifications.
It's a parts Sandra FBI, which again is compatible with existing Apache Cassandra libraries on its support. Native FBI, which you can quickly with sequel like query language
in terms of some off the car functionalities off Azur Cosmos TV I mentioned Hey there, it's got global replication. It's actually just don't care application. You simply click a button on the mouth on replication or cost transparently in the background to that region. And what that means is that you can put
not only your applications close to the users that are using the applications.
You can put your database very close to the users that I use in the application.
It has very inconsistency level would talk about that a little bit in a minute because I joke was mostly be support multi master, right? So and we have to have a wayto handle and consistency, and it has different degrees of consistency levels that we can configure.
It supports. It has low latency by its it can. It serves 10 lesson 10 minutes account for Reed operations and less than 15 minutes. Account right
request for night night be sent off the operations.
It's got elastic scale out. True food can scale from 10 toe. Undress off millions of requests for seconds across multiple regions.
I mentioned this a little bit earlier, but what's touching on again? Because when it comes to Cosmos TB,
we can migrate existing monk body be databases and Cassandra databases towards your cosmos. TB on. We do not need to transfer mount up our applications in any way because with the same libraries on SD case, we can communicate without your cost. Mostly, be really happy. High is in the same way up for the calls
that the application's already use.
Now let's talk a little bit about this consistency levels.
So because, as your cosmos TB supports Marty Master right, what it means is there has to be a way for us to specify
for many Nazi quote little business. They have only two degrees. I can configure it a strong consistency, which means
the whenever there's a right operation that committed, it has to be replicated across all the different masters before an acknowledgement is supplied.
All they have the eventual consistency which is where you can, You can say, Yeah, I don't care whether that I get stay as long as it gets there now Venture consistency is great for performance, but it's not good for availability. Because if they were to be a failure on the single master that probably
one of the masters that holds a ride up right operation before it's replicated.
Then Dad, it is lost.
Where's the strong consistency is great for availability but not great for performance. What, as your cause mostly be does await. Different is that it doesn't only have strong and eventual, but it has, like all these varying degrees in the middle, not all applications needs to go for eventual off a strong
you can either ghost pounded state less session or consistent. Terrific. So there's a whole spectrum
that we can select from when it comes to consistency levels.
Some more information about this consistency level. So we have strong the bounded state last session, the consistent prefix and the eventual strong and eventual, we explained earlier.
When it comes to bounded statelets else downstate less. What that means is we can sort off set limits to say make sure that the replication is consistent within this amount of time. In other, what's where What we're saying is we're saying
go ahead and do the replication, but you can only be still for this amount of time out for this amount off, off, lug behind in terms of consistency.
So we can specify that when it comes to session, which is what what many applications. Actually they need right when it comes to session consistency. If to say, as long as the request is coming from the same client hall, read and write, operation should be consistent across the user session, and for many applications this is sufficient.
And then you have the consistent perfect, which is sort of like
went Just make the data will be replicated
if at some point it as like loose consistency by the fan is when the doctor gets replicated, they've got to be replicated in the right order.
So you have that balance between performance and availability, which you have to select the one that
fit well with the application, the good finish. You have the ability to do this election.
Now let's talk a little bit about the practical side off working with azure cosmos TB
when working with Jacques Cosmos, baby, the first thing that we need to create a something called and as your cosmos TB account
No. Where we create as your cosmos TB account within our account, we can then create databases. So if you're thinking about it in terms of finishing out database that are being sense off my county's oh, you have your server and then after that, you have your databases within a seven, you can have multiple.
Then with dinner database, we can have something called a container.
Now, when it comes to the container,
it varies depending on the A P I selection that we choose when we're creating the cosmos TB account. So, for example, the container can be a collection. It can be a graph. It can be a table. If you select
Cassandra or table FBI, for example, container will be reflected as a table. If you selected Gremlin, you be reflected as a graph. It selected mongo D B. It will be reflected as a collection it selected cycle a p I. It will be reflected as a container
on, then under the containers. Then we can have our high temps. So think about like the table as sort of like the
like like containers off like a table within the relational database and the high temps of sort of like develops and the columns within that. So, and that's what you're talking about. You're talking about the actual document or the high times that are. Start with Dean our containers on. Also, we can have teams like Strobe Start for CDOs, you said, is that define functions and entry guests and conflict.
I know those are the heightens,
so I'll post this video right here.
Ah, nde when I get to the next video, our carry on talking about azure Cosmos tibia, and I'll give my information around things like petition in Frazier Cosmos TB on. Then we'll get into the design decisions
off Advocaat's mostly Be who also cover cost mostly be migration on my great and data into Cosmos TV. Thanks for watching this video and see you in the next lesson