A few weeks ago we announced our partnership with Google Cloud to jointly develop secure solutions for customers in the areas of Artificial Intelligence, Secure Hybrid Cloud, Data Analytics and Digital Workplace.
Yesterday we’ve had the chance to attend the Google Cloud Onboard event which is a training day on Google Cloud Platform (GCP) fundamentals hosted in Zurich.
In this article we will try to cover the highlights on the training program, consisting of the main offerings of Google Cloud Platform.
Google stands on two major pillars:
- the infrastructure driven by the technical power
- the information needed by the business
The continuous advancement in the era of global business and the expansion of the amount of data is requiring a constant improvement of the global infrastructure. The cloud infrastructure is the optimal solution to balance the needs of the business against the increase of data and information.
Google Cloud Platform offers efficiency and availability by having Regions and Zones all around the globe. With more than 10 regions and 60 zones, the service insures high performance to the business. Other than the availability the platform allows companies to outsource their security to the cloud in a way that ensures business continuity and prevents disruptions or attacks.
— Atos_CH (@Atos_CH) 8 maggio 2018
The training covered a detailed overview of the fundamentals of the Cloud Platform which consisted in four main solutions: Compute, Storage, Big Data and Machine Learning.
All solutions embrace opensource as a main strategy to improve sustainability and allow business to be flexible and customize their functionalities according to their needs.
The solutions mentioned include the following services:
1- Compute: Compute engine, Kubernetes Engine, App Engine and Cloud Functions
2- Storage: Bigtable, Cloud storage, SQL, Cloud Spanner and Cloud Datastore
3- Big Data: BigQuery, Pub/Sub, Dataproc and Datalab
4- Machine Learning: Natural language API, Vision API, Machine Learning, Speech API and Translate API.
This section only illustrates some of the highlights of the training day. Mostly all of the platform aspects are very interesting and detailed information could be found on the official Documentation site.
For each solution we chose some of the offerings and services that we consider appealing for the reader:
Virtual Machines and Networks
The Networks in GCP act as virtual containers: each network could be boxed as a single container across multiple regions with same IP and Gateway details. This allows global organizations to have their own networks across the globe with high performance based on geographic allocation. However, companies have also the flexibility to interconnect their business using their own networks.
Like most cloud services do, Google offers Remote Virtual Machines hosted in the cloud and managed by their own network and security. You can create or import your Virtual Machines by accessing the Compute section in your platform. Importing preexisting virtual machines is integrated and can be done with few clicks to optimize the migration process.
Containers and Kubernetes
Using containers frees the business needs from the boundaries set by the infrastructure, allowing the interaction with virtual services without duplicating the operating system layers. These standalone, lightweight and executable containers nowadays rely on Docker, which is the only Enterprise-ready container platform.
To manage, deploy and scale the containers you can use Kubernetes Engine that consists in multiple machines grouped together to form a container cluster. Each of these clusters could be orchestrated by Kubernetes cluster management system.
— Atos_CH (@Atos_CH) 8 maggio 2018
There are four main categories in the Storage portfolio in GCP:
Cloud SQL: a fully-managed database service that supports MySQL and PostgreSQL. It’s easy to set up, maintain and manage.
Cloud Spanner: a large-scale powerful database that combines relational vertical scale and non-relational horizontal scale with high performance.
Non-relational (Flat NoSQL)
Cloud Datastore: a highly-scalable NoSQL database which is integrated for App Engine Applications such as Mobile and Web.
Cloud Bigtable: a Big Data database service for heavy read/write events such as analytical data (finance and IoT).
This service allows the storage of structured and unstructured binary or object data (images and media files).
BigQuery: a serverless highly scalable data warehouse for massively large datasets.
3- Machine Learning
Machine learning is a field of Artificial Intelligence that solves problems without explicitly codifying the solution. The systems involved improve and learn to improve themselves over time.
With GCP you can use pre-trained models and services and connect them to your needs. Those models can later be trained to improve and predict corporate outputs.
All of the processes are in a serverless environment that responds in real time with high availability.
Some of the fully-trained models as REST API
Vision API: It analyzes, understands and classifies the content of images. It could detect objects and faces within an image.
Speech Recognition API: It converts the audio to text, or the other way around, by applying powerful neural network models. The text tonality is “intelligent” and not static, it learns and adapts to the language spoken.
Train pre-implemented models
Cloud AutoML: It enables developers with limited machine learning expertise to train models. Example in the training was AutoML Vision (Alpha version) which was trained to recognize the type of clouds based on pictures uploaded to the system.
Build custom models
ML Engine and TensorFlow: It allows custom development of your own models into Google Cloud Platform.
4- Big Data
Big Data can be fully managed in GCP with a serverless approach, that means that the backend architecture is handled by the service provider.
BigQuery: As an example in this overview training the instructor triggered a query that fetched a huge amount of data. The result was given within 1 seconds with cached query result and less than 10 seconds with query caching disabled.
BigQuery is a cloud Data Warehouse that supports immense amount of business data.
Dataproc: This service is to use clusters of Apache Spark and Hadoop on GCP in a scalable cost effective way.
The Google Cloud OnBoard day was a very informative day about the Platform fundamentals. Of course, further training is needed to gain a better knowledge about the different solutions.
A very nice thing about this event is that Google gives this training for free and provides participants with free additional trainings to expand their knowledge, e.g. Kubernetes in Coursera.
Everyone could simply log in Google Cloud Platform and try the platform for free to evaluate the products and services.
Looking forward to the new solutions this platform and our partnership will allow to create!