Looking for an effective platform to deploy your machine learning models. Are you stuck with exploring Cloud API? Here is the easy and efficient way to store and run your applications by – Pay-per-pod Kubernetes Cloud! The First-ever Kubernetes Cloud.

You might also like ParamSiddhi AI, Microsoft AI Classroom, and Quantum Machine Learning.

For more such topics – Click Here.

If you will go around the lifecycle of your application, you need to manage cloud instances, virtual machines, Kubernetes clusters, etc. and this might be cumbersome. If you don’t have a strong technical hand in your group or you are a beginner in such a technological stack, then it is a major issue to consider.

Here, we have a solution for you! Puzl provides the most efficient way to handle such a situation most easily and efficiently so that you can run your applications in just one click! All you need is an account and your application, with this you are ready to go!

Puzl provides the user with full access to the Kubernetes namespaces where he/she can have direct control over it. There is no need for the user to manage all the things like Virtual Machines, Cloud Instances, or even the Kubernetes Clusters. You can manage your Kubernetes namespace only as you have direct access to Kubernetes API.

One of the great things about Puzl is that you pay-per-pod Kubernetes Cloud and that too depends upon what your Kubernetes namespace contains. If it is empty then you don’t need to pay anything, and that’s great! The resources which the running pods use and the data in your Kubernetes namespace comprise the total amount that you need to pay!

By using cloud services directly, you would have to pay in tons as the users grow! But don’t worry, puzl has brought a solution for such a situation, where you can run your applications by using GitHub repositories and Docker Hub. You can upload your application/model from the GitHub repositories or Docker Hub to puzl, in the Kubernetes pod that you have got created and take the benefit of it!

By using Puzl it is easy to allocate resources and resource management is done efficiently. Its basic advantage is that it is flexible to use and run the applications. As you don’t need to manage the Kubernetes Cluster, it is easy for you to use Puzl if you are a beginner. The main concept is that you don’t need to manage the cloud instances or there is no need to deploy whole Kubernetes clusters, all you need is to scale up your resources as per your requirement and you can start learning Kubernetes on Puzl also, in case you aren’t familiar with the concept of Kubernetes. Within a few seconds or minutes, you are ready to go!

It provides you with all the essential services running, service policies, security policies, network policies, and so on. It helps the user to manage and organize his/her resources efficiently so that he/she would utilize them in the best possible way. Here one can run the pods directly once a resource has requested it. Once a user joins Puzl, the Kubernetes Cluster and namespace will be generated within a few seconds and under the assistance of the Puzl help center which will solve the maximum of your doubts in their FAQ section.

Kubernetes

Kubernetes is an open-source platform that provides the option to manage workloads and provides the user with various services, which help in better organization of resources and automation.

Basic Information

Pod

In a specific Kubernetes cluster, a pod is a group of containers over the same network which share the same resources and storage. The containers can communicate with other containers that are within a single pod so they share a common network but are isolated from the container within other pods.

Pay-Per-Pod

Pay-per-pod means that the user won’t be charged until he/she creates a pod for availing certain resources and the minimum resource for one pod is 0.1vCPU and 128 MB RAM.

It uses professional server platforms with a high-bandwidth local network, which are placed in Tier III data centers providing high power supply and reservation of power facilities.

Here you don’t require the cloud or bucket concept, using your Docker images or git repositories, one can run/deploy his/her applications by using the Kubernetes cluster directly. For user support and overview, it provides various repositories, applications, and images like TensorFlow, SSH server, Jupyter Notebook, etc. Here one can easily deploy his/her application without taking care of infrastructure or virtual machines. One just needs access to his/her Kubernetes namespace and its basic idea, which is provided by the Puzl itself, and then the person is ready to go.

Here, users are provided with separate Docker images on which they have to run their application and each user has access to a separate namespace. Namespace refers to a mechanism in which the resources of a Kubernetes cluster are divided so that multiple users can share those resources which reside within a single cluster. All the resources and networks of a particular namespace in the Kubernetes cluster are isolated from the resources and networks of other namespaces in the same or different Kubernetes clusters. Hence it is easy to use and comes with high security and privacy options. It is efficient to use and comes with a user-friendly environment.

How to tackle Technical Issues?

The Kubernetes API page in the dashboard comes with all the instructions that you need to follow while creating a Kubernetes cluster denoted by kubectl. You need to ensure that your Pod is running before beginning further processing. All the codes along with the steps in detail will be provided to the user on the Kubernetes API page which is easy to follow along.

The user can add the SSH keys after pod creation also, it may take a few minutes. You can execute a pod by cloning a git repository as well. Make sure that the repository has a unique name but if the name exists previously in the folder then make a pull command. For a better experience, you can change your target folder also. Once you have created a pod using the git repository, you can also add packages. You can also get access to private Docker images from the Docker Hub only after creating a pod.

To maintain security and privacy, cloud bounds connection to a few ports: 25, 587, 465, to minimize or reduce the risk of the privacy policy.

Root User

The root user can be used in any container and runtime system package installation is preferred so that each process is repeated at the start of each pod. The directory where the git repo was cloned is the current directory. All the containers that we create in a pod are created with a user, i.e., with a particular Kubernetes namespace provided.

Restrictions

Prefixes like *.puzl are reserved initially and their proper working cannot be guaranteed. The user cannot create envs with names starting with _PUZL as these are reserved. All the things should be checked and configured before as if the criteria don’t match or have any other faulty value then the pod with that particular content can be terminated.

Payment Mechanism

The minimum balance top-up is only 10 euros and they do not require the user to link a card with their account. In the future, they might provide a completely free account with a limited set of resources, so that everyone can have the benefit.

The Debit is calculated per second so you need to maintain and manage it. The billing mechanism takes care of the pods created and the resources allocated to each pod in the Kubernetes cluster. In the case of pod failure or damage, specific care is taken for the bill generated. If the pods restart again, all the resources that were requested by the pod previously will be allocated to it.

Once a pod requests some resources, they are allotted to the specific pods which have been requested for them. The information and all the details about the resources are mentioned in the dashboard of the user. Once the resources are allocated and pods are filled with resources, the billing of these pods starts as per the number of resources allocated to them.

The user should take care of the following points:

• As soon as the credit becomes less than 0, all the services would be deleted. At the same time, the volume with the data remains available, the user can top up the balance, create an SSH Pod for access, and download the work results.
• When the balance falls below -20 euros, the volumes are deleted.

The user receives an alert through email when the balance drops to the threshold.

Puzl provides a much better alternative so that the services won’t be stopped and the user account won’t be terminated if in case he/she wasn’t able to manage the account. It is done using the Pre-paid model, it allows the users to use the services just after they have registered, without any time delay and that too irrespective of any additional account checks.

About Resources

The user can upload or download the data by creating a pod with SSH, this will allow data transfer. After the user has filled the volume for a particular pod, he can remove the SSH pod and link it to another pod to fill the volume, i.e., upload or download the data.

Shared or Exclusive CPU Cores?

Puzl comes with both options. You can choose shared CPU cores for high performance but due to high L3 cache usage, the performance of shared CPU cores may drop, and hence in such cases, exclusive CPU cores are recommended.

There are specific settings and criteria that one needs to have a look over and all these things are provided in the Help section, in detail so that the user can follow along with the instructions and do all the set-up requirements so that he can avail maximum benefit of it,

The specifications, features, and quality of services of the pods depend on the requests that the pods made, the associated limits, and the quality of resources allocated to the pods. These factors, though, are varying for different pods but they provide an overview of the functionalities of the pods and the services to which the pod has access.

For RAM and CPU there are specific containers that specify the limits of volume that they can have. Each pod is provided with the same amount of memory limit and memory request. In the same way, each pod is provided with the same amount of CPU limit and CPU request.

Generally, pods have a memory limit greater than the memory request and a CPU limit greater than CPU requests and if a pod exceeds the limit of memory request then it can be terminated forcefully from the specific Kubernetes cluster.

Different terms define different status terms of the pods:

·   Requested – The pod has requested some service or resource which is being processed by the Kubernetes.

·   Pending – The status of the pod which tells that the pod is waiting for its request to be processed.

·   Scheduled – The resources that were requested are now allocated and at least one pod container is ready to extract its Docker image.

·   Running – This shows that the pod is performing its operation without any obstacle.

·   Succeeded – The pods have performed their operation successfully and the pod’s container has finished their job with exit code 0.

·   Failed – The exit code of at least one pod is not 0 which means that the pod hasn’t performed its operation successfully.

·   Terminating – The pod has requested that it is to be deleted.

·   Terminated – Pod is deleted from the specific Kubernetes cluster.

Just like pod status keywords, we have volume status keywords that denote the different status conditions of the volumes.

·   Requested – The request is received by the Kubernetes server and is being processed.

·   Pending – It is waiting to be processed by Kubernetes.

·   Running – It shows the volume is performing its operation without any issues.

·   Terminating – That volume has requested that it has to be terminated.

·   Terminated – Volume is deleted from the specific Kubernetes cluster.

Container File System: container – fs

There are several specific container operations that the pod needs to perform and the data associated with such operations is stored in the Container File System. The specific limit of the container file system is 5GB which can’t be increased. The Container File System is created automatically when the pod is created and gets deleted automatically with the deletion or termination of the pod with which it is associated.

It is dependent on the pod, i.e., its life cycle relies on the life cycle of the pod and is used to manage and store the data of the container operations in the most organized and efficient manner.

Benefits of Puzl

As it provides the users with the facility to upload their applications through Git repositories or from Docker Hub that needs to be managed, it comes with the option of instant start, where a new user with low technical knowledge can also deploy his application most efficiently.

The bill generated is quite low, the user needs to pay only for the resource that is requested by the pod created by the user. There are no separate charges for cluster maintenance.

With the Kubernetes to run the applications directly, it has provided an easy and time-saving platform where one can deploy his application within a few minutes.

It provides various services that will speed up your flow and provide a great kick start in the market. These services include SSH server, Gitlab CI Runner, Jupyter Notebook, and various new features and services are coming soon which include H₂O.ai, PostgreSQL, Determined AI, Redis, Jupyter Hub, Drone CI, MongoDB, Airflow, etc.

The new services that have paced up the market flow include efficient GPUs with efficient and cheaper features and they have provided it brief detail as well.

For a new user the specifications or limits of the Kubernetes namespace allotted are:

·   Pods – Max : 20, Available : 20

·   Volumes – Max : 4096 GB, Available : 4096 GB

·   Ports – Max : 30, Available : 30

·   RAM – Max : 384 GB, Available : 384 GB

·   vCPU – Max : 96 vCPU, Available : 96 vCPU

·   GPU – Max : 10, Available : 10

The dashboard lists all the details like maximum resources, available resources, and used resources. The used resources are 0 for a new user. It gives the accessibility to create or deploy an application, create a volume or pod.

The user can have a look over his workspace containing pods, volumes, and applications, his credit balance, Kubernetes containing information and details about API and dashboard,

The dashboard provides a brief look into the user account where the user can maintain and check his profile, be updated about billing and security-related information. There is a new feature – Marketplace, where the user can have a look over all the features, resources, and services that are available or are coming soon.

They have provided with the resources which a user can refer to his benefit. These resources are provided via GitHub and Docker Hub for the benefit and learning of the users, especially of new users.

Along with all these features, services, and resources, they have provided the user with the Help option, which the user can refer to in case of any query or doubt. There is a pricing section also, which gives a brief detail about the pricing of each service and resource depending on their specification, time limit, location, and other dependencies.

They have a specific section that provides all the billing-related information like billing history, billing details, balance, and details about the date of usage. It helps to maintain transparency among the users and the company.

Overall, puzl provides a cheap, efficient, resourceful, and easy environment where one can store and run his application.

For more such topics – Click Here