A Walkthrough of public Cloud vendors (AWS vs GCP vs Azure) and their AI Capabilities
A presentation at AI Helsinki March 2019 in in Helsinki, Finland by Bruno Amaro Almeida
A walkthrough of Public Cloud vendors (AWS vs GCP vs Azure) and their AI capabilities AI Helsinki (March 2019) 27.03.2019 Bruno Almeida BERLIN · HELSINKI · LONDON · MUNICH · OSLO · STOCKHOLM · TAMPERE
2nd level subsection title Subtitle
$ whoami Principal Architect & Technology Advisor @ Futurice ! native, based in ” Architecture, Cloud, Security, DevOps & AI
coffee, music, travel & indoor climbing
Reach out on: @bruno_amaro BERLIN · HELSINKI · LONDON · MUNICH · OSLO · STOCKHOLM · TAMPERE @brunoamaroalmeida
Public Cloud: AWS vs GCP vs Azure What are our options? BERLIN · HELSINKI · LONDON · MUNICH · OSLO · STOCKHOLM · TAMPERE
Public Cloud: Global Landscape Public Cloud Global Market Share (CSA - 2017) Gartner Magic Quadrant for Cloud 2018
Overview • Market leader year-over-year • Extreme Customer centric • Really good at providing services that do the heavy lifting of common things you can build by yourself but you don’t really have (or want) to • Very stable services • Built from Ops towards Devs point of view
Overview ● ● Gaining market share very rapidly. Good offering and value proposition for hybrid environments (on-prem <-> cloud or cloud <-> cloud ) ● Europe North == Ireland / UK / Germany. Currently no region in the Nordics (but Oslo coming soon). A region has one or more AZs. Paired regions concept. ● Things-that-look-quite-interesting: ● ● Service Fabric ● Azure DevOps ● Resource Manager Template (IAAS) ● PowerBI
Overview ● ● ● ● Strong focus on Innovation and providing services you can’t reasonably build by yourself Built from a Developer point of view, designed with a Site Reliability Engineering mindset Some services released as OpenSource (e.g. Kubernetes -> GKS), others fully commercialized (e.g. Dremel -> BigQuery) Things-that-look-quite-interesting: ● ● ● ● Pub/Sub Service A lot of services were designed to be global ( internet scale) by default Very Interesting AI/ML offering and capabilities Datacenter in Finland
AWS vs GCP vs Azure: Core Building Blocks Compute Network Security & Identity Storage • AWS EC2 • AWS VPC • AWS EBS • AWS IAM • AWS ECS / EKS / Fargate • AWS Route 53 • AWS S3 • AWS KMS / CloudHSM • AWS Lambda • AWS Elastic Load Balancing • AWS EFS • AWS Inspector / Advisor / GuardDuty / Shield • AWS Elastic Beanstalk • AWS CloudFront • Google Compute Engine • Google Cloud Virtual Network • Google Persistent Disk • Google Cloud IAM • Google Container Engine / GKE • Google Cloud DNS • Google Cloud Storage • Google Cloud KMS / Cloud HSM • Google Cloud Functions • Google Cloud Load Balancing • Google Cloud File Store • Google Cloud Security Scanner • Google App Engine • Google Cloud CDN • Azure Virtual Machines • Azure Virtual Network • Azure Disk Storage • Azure Active Directory • Azure Containers / AKS / Service Fabric • Azure DNS • Azure Blog Storage • Azure Key Vault / Dedicated HSM • Azure Functions • Azure Load Balancer • Azure File Storage • • Azure App Service • Azure CDN Azure Sentinel / Security Center / DDoS Protection
Public Cloud: AI & Analytics Capabilities What do they offer? How do they differ? BERLIN · HELSINKI · LONDON · MUNICH · OSLO · STOCKHOLM · TAMPERE
AI & Analytics Capabilities Data Engineering (ingest, prepare, transform, analyze) AI/ML Platform (build, train, deploy) AI/ML API’s (pre-trained models, serverless, out of the box)
AWS vs GCP vs Azure: Data Engineering Ingest • ETL AWS Kinesis • Google Pub/Sub • Azure Event Hubs • • • AWS Glue / EMR Google Dataflow / DataProc Azure DataFactory / DataBricks • Data Warehouse Raw Storage • • AWS S3 Google Cloud Storage Azure Data Lake Storage • • • Machine Learning AWS Redshift Google Cloud BigQuery Azure SQL Data Warehouse AWS SageMarker • • Analytics / BI Google Cloud Datalab • AWS QuickSight • • Google Cloud Data Studio Azure ML Studio / Workbench • Power BI
AWS vs GCP vs Azure: AI/ML Platform AWS Sagemaker Google Cloud ML Engine • Amazon SageMaker automatically configures and optimizes TensorFlow, Apache MXNet, PyTorch, Chainer, Scikit-learn, SparkML, Horovod, Keras, and Gluon. • Commonly used machine learning algorithms are built-in and tuned for scale, speed, and accuracy with over a hundred additional pretrained models and algorithms available in AWS Marketplace. • You can also bring any other algorithm or framework by building it into a Docker container. Source: AWS • Cloud Machine Learning Engine: Massively scalable managed service for training ML models & making predictions. • Enables apps/devs to use Tensorflow on datasets of any size • Supports online & batch predictions, prioritising latency (online) & job time (batch) • Or download models & make predictions anywhere: desktop, mobile, own servers Azure ML Studio • Azure Machine Learning: End to End Data Science Solution • Uses PyTorch, Tensorflow and Keras • Multiple Components: • ML Workbench • ML Experimentation Service • ML Model Management Service • ML Libraries for Spark • Visual Studio Code tools for AI • HyperTune automatically tunes model hyper parameters to avoid manual tweaking Source: Google Cloud Source: Azure
AWS vs GCP vs Azure: AI/ML API’s AI/ML Service APIs AI/ML Service APIs AI/ML Service APIs • AWS Lex • Google Dialogflow • Azure Bot Service • AWS Rekognition • Google Vision API • Azure Vision • AWS Translate • Google Text-to-Speech API (ASR) • AWS Polly (TTS) • Google Speech-to-Text API • AWS Transcribe (ASR) • Google Natural Language API (NPL) • Azure Speech • Translator Speech API, Bing Speech API • In preview: Speaker Recognition API, Custom Speech Service • AWS Textract (OCR) • Google Translation API • Azure Knowledge • AWS Comprehend (NPL) • Google Video Intelligence API • AWS Forecast (Time-series forecast) • Google Inference API (Time-series forecast) • Azure Search • Bing News/Web/Image/Video/Custom Search • Google Job Discovery • Google Cloud Genomics (Store and process genomes and related experiments ) Source: AWS Source: Google Cloud • Azure Language • Language Understanding (LUIS), Bing Spell Check, Text Analytics, Translator Text API Source: Azure
Thank you! Kiitos! Danke! Tack! Bruno Almeida PRINCIPAL ARCHITECT & TECHNOLOGY ADVISOR Cloud, CyberSecurity, DevOps, Data Engineering & AI Reach out on: @bruno_amaro @brunoamaroalmeida BERLIN · HELSINKI · LONDON · MUNICH · OSLO · STOCKHOLM · TAMPERE
Buzz and feedback
Here’s what was said about this presentation on social media.
