Explore the basic technologies in public cloud platform: Amazon Web Services (AWS), Microsoft Azure, and Google Cloud Platform (GCP).
Compute
#- At on-prem, or traditional data center, Compute is referring to the physical server.
- At cloud, Compute is referring to
virtual machine
or container
or serverless
. - typically,
virtual machine
isn’t the go-to technology to use nowadays. serverless
solution is easier to be used (called endpoint).
Compute Technologies | AWS | Azure | GCP |
---|
Serverless | Lambda | Azure Functions | Google Cloud Functions |
Virtual Machine | EC2 | Virtual Machine | Compute Engine |
Hosted Container | Fargate | Container Instances | Cloud Run |
Container Orchestration | EKS | AKS | Kubernetes Engine |
Storage
#- For storing data as object.
- Can apply labels and filters to the object.
- typically used for long-term storage, like PDF and word files (unstructure data).
Object Storage | AWS | Azure | GCP |
---|
Storage | S3 | Blob Storage | Cloud Storage |
Database
#- For structure data which allows for querying and complex transaction.
- Optimized for frequent access.
- More specialized DB such as knowledge graphs, vector DB.
Database | AWS | Azure | GCP |
---|
Relational DB | MySQL/PostgreSQL | MySQL/PostgreSQL | MySQL/PostgreSQL |
NoSQL | DynamoDB | CosmosDB | Firestore/datastore |
Data warehouse | RedShift | Synapse Analytics | BigQuery |
AL/ML
#- Solutions for vision-focused, generative AI, etc.
- Still new.
AI/ML | AWS | Azure | GCP |
---|
Pre-built AI | Amazon Rekognition, Amazon Translate, Amazon Comprehend | Azure Cognitive, Azure Machine Learning | Google Cloud Vision AI, Google Cloud Translation, Google Cloud Natural Language |
ML | SageMaker | Azure Machine Learning | Vertext AI |
AutoML Tools | SageMaker Autopilot, Amazon Kendra | Azure Automated ML | AutoML Vision, AutoML Natural Language |