Nebula Graph NebulaGraph for Beginners, 9 Aug 2021

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Nebula for Beginners - Operator explained (CN Version & EN Subtitle)

Description

This video explained in detail how to use Nebula Operator to deploy Nebula Graph on K8s. Besides, what's the difference between deploying Nebula Graph on Bara Metal and Nebula Operator. Welcome to discuss on our Slack: nebulagraph.slack.com
Tips: if you want to see English subtitle, please switch the subtitles to English in the settings.

A

Hi ~, hello, everyone, I'm thinking this is as hera, hera, hi, ~ hello,. Everyone met again.

B

Do good to our detailed understanding of what this week nebula operator, please think, first, to tell you about as.

A

Good hera, then we start with what is a kubernetes (k8s), the operator to begin.

A

In k8s we know we need to maintain a wide variety of human resources and multi-level.

A

Such as life cycle management do when we need a variety of resources at all levels of all sorts of resources to do operating.

A

K8S, a control loop, which has helped us in accordance with a change in status of this resource, to help us apply the appropriate operation and maintenance of operating.

A

In general for complex applications that we do not only need to care so much sophisticated resources, while we still have to application level to do some corresponding operations.

A

Such as nebula graph, we need to do some balance, data,, etc. operations that complicated operations for the operation and maintenance. Personnel, then even in the k8s, still is not easy.

A

So that it produces in the ecological k8s such a pattern of a group called operator.

A

Operator with it, we can put some of the operation and maintenance of complex applications do more work and more abstract automate. Some of the fields we passed, some of the resources from the definition of.

A

What can be abstracted from the upper application and let some operations change the operator to help us to do some of the state of the underlying resources.

A

We can to expand k8s the control loop and hook it to be able to deal with some of this resource mobilization, corresponding complex applications above.

A

It,, while some of us will be in accordance operator change conditions to automatically help us to call the appropriate application-level api is not just k8s api. That.

A

Makes our operation and maintenance operations more automation, effort, less error-prone.

A

What is that nebula operator it natural that it is what we.

A

Specialize for nebula graph.

A

After kubernetes of operator with it, we expanded it, a logical control plane on k8s can be very, very easy to do easier deployment easier to do. The installation.

A

Configuration nebula graph, re-configuration, (second configuration) or is scaling capacity either upgrade or downgrade or fail over backup and recovery operations.

A

So complex and so nebula graph of operator is such a.

B

Good thing, after understanding that you can tell us something about (specific) how to use nebula graph do.

A

Good hera, that how to use it is actually very simple.

A

We just need to have it installed on some k8s environment, (nebula) operator itself,.

A

To achieve this, expansion and above to operate as defined above can crd specifically, in order to make the operation.

A

That it is what we have before we go to such a k8s installation I want some of it will depend. Nebula operator now rely on these three components: (certmanager openkruise, helm),.

A

We install it after it directly through helm.

A

A key to our nebula operator extension pull up that after pulling up easier our for the operation of the entire cluster,. It would only need to focus on such a crd, some of the following parameters.

A

At this level do operate on, we can do some operation and maintenance work. for example,. If we want to expand and shrink graphd,, we notice that the graph replicas is now 1.

A

Let's change it to 3, for example. apply this yaml file.. The operator will automatically compare this diff and then help us automatically. to expand the capacity, extra 2 graphd.

A

For upgrading and downgrading, etc., and a series of other complex applications, including automatic, fail over. for.

A

Example, we can artificially fake some failures in the production environment.. We can kill 1 graphd pod.

A

Operator. this distinction automatically detected. It will automatically help us to restore.

A

And then create a new repair such a pod, so this is how to use a.

B

Good introduction operator operator, it sounds very convenient that the production environment, if used in binary deployment,.

B

Or deployment nebula operator, what is the difference between what they.

A

Actually use in the case of a binary nature of its underlying applications are running nebula graph.

A

That in binary or bare metal, we call it it has. An important difference is that it is not a container so that the deployment of some of the ease of use of a lot of k8s or container to bring it. We lost the.

A

Other aspects of the configuration is.

A

We've just mentioned with crd, to abstract the fact that our entire nebulacluster.

A

Configuration the format has also changed, because we passed a mapping of a key value of the config in k8s and mapped it to some configuration files.

A

Of nebula graph,, so its format is a bit different.. One of us is to manage the configuration file and the other is to manage ours. on yaml key value of children. (Pair).

A

There is also self-healing upgrade backup and so on, then.

A

If it is bare metal, (binary) deployment,, then we need to write their own scripts to control.

A

Then we need to do some call other scripts, but with the operator,, then we only need to go inside the operator. This yaml, given some resources, can be a.

B

Good explanation for the xiexie si: there is also a possibility that we would be more concerned about the performance difference between what they do.

A

First is to say, we k8s, which is capacitive deployment on the performance of.

A

That fact,, we know that container brings loss is very small. It does not like some virtualization hypervisor.

A

But is not without influence, because if we go by the operator,, then we are actually installed yet some of the workload control surfaces installed.

A

That some of these, such as operator or even controler, we took some sidecar. Some of these things were actually consuming cpu and memory resources that it is part of the price of course,, but.

A

It brings the convenience of our entire dataplan. Our level of data loss is actually very little that can be considered.

A

It is worth mentioning that the point is that nebula graph us is based on the underlying is rocksdb. It is actually based on the assumption of ssd a design.

A

If some of our k8s cloud environment it uses pv it's those volume, if not local ssd. If such it is a net disk,, then.

A

In theory that it was not us on bare metal with local ssd can have the same.

A

Performance, of course, we actually use the k8s is still possible to achieve the same performance by using local pv manner. So this is worth noting.

B

Thank you for your nice little good to explain that if just want to try nebula operator, what's up with the fast method, well,.

A

We know that, although the operation and maintenance of our operating day1 day2 operator has become very easy to very easy.

A

But the actual the operator increased complexity of deploying its own, it did not seem so easy,, especially, it depends on a number of things on the k8s for the novice.

A

As well as k8s, if want a quick early adopters,, then I suggest that you can refer to our official one article.

A

The first is that the above article is a detailed description of the operator. Is our development qiaolei his students, a good introduction, followed by a second article,? It is zhang, wei.

A

A classmate made a presentation to get started. You can follow this article (step) on ali cloud, convenient to deploy a third nebula operator environment.

A

I would like to introduce,, then you can help us solve this problem. Is, is my personal project. I wrote a small advantage, (operator) kind.

A

Of such a on the inside of a k8s interest group called k8s in docker. I can make use of this little project.

A

You can, for example, help you automatically in a deployment docker docker deployment of a key on a single 8g memory on a google cloud, a key k8s.

A

And all dependencies on deployment include helm, certmanager and local volu me.

A

And then, on top of that, to help us deploy (nebula) operator and go up the deployment of this in a cluster (around a bit). It is a good way to help you a fast, a key early adopters, operator's.

B

Good that, I just add something to think said that the operator kind can be thought of as a personal website. Siwei.Io go inside to find the resources.

B

You can go to try and have any questions or doubts can communicate with us in a micro-channel in the forum,. We will reply to you.

A

That this program will I'm here, thank you,. Think for goodbye,. Thank you. Bye.