Migrating Existing Open Source Machine Learning to Azure
Download as PPTX, PDF6 likes14,845 views
The document outlines the use of Microsoft Visual Studio tools for AI and Azure Batch AI for streamlined computational tasks across different operating systems. It highlights the flexibility and scalability of cloud computing, allowing custom cluster creation for specific jobs, leading to cost savings and efficient resource utilization. Additionally, it mentions practical applications, such as utilizing Apache Spark for analytics and machine learning tasks.
![Visual Studio [Code] Tools for AI
VS & VS Code extensions to
streamline computations in
servers, Azure ML, Batch AI, …
End to end development
environment, from new project
through training
Support for remote training & job
management
On top of all of the goodness of
VS (Python, Jupyter, Git, etc)
THR3129 Getting Started with Visual Studio Tools for AI, Chris Lauren](https://image.slidesharecdn.com/brk3709smith-180508160604/85/Migrating-Existing-Open-Source-Machine-Learning-to-Azure-9-320.jpg)
![Series RAM vCPU GPU Approx Cost
Standard_B1s 1 Gb 1 None Free [*]
DS3_v2 14Gb 4 None $0.23 / hr
DS4_v2 28Gb 8 None $0.46 / hr
A8v2 16Gb 8 None $0.82 / hr
Standard_NC6 56 Gb 6 0.5 NV Tesla K80 $0.93 / hr
Standard_ND6s 112 Gb 6 1x Tesla P40 $2.14 / hr
[*] Not recommended: Standard_B1s (free, but too small to be useful)](https://image.slidesharecdn.com/brk3709smith-180508160604/85/Migrating-Existing-Open-Source-Machine-Learning-to-Azure-15-320.jpg)
Migrating Existing Open Source Machine Learning to Azure
- 7. Replicable and scriptable Consistent syntax on Windows (cmd / Powershell), Mac, Linux, WSL cda.ms/sH
- 9. Visual Studio [Code] Tools for AI VS & VS Code extensions to streamline computations in servers, Azure ML, Batch AI, … End to end development environment, from new project through training Support for remote training & job management On top of all of the goodness of VS (Python, Jupyter, Git, etc) THR3129 Getting Started with Visual Studio Tools for AI, Chris Lauren
- 12. cda.ms/sN
- 14. • Local tools • Local Debug • Faster experimentation Single VM Development • Larger VMs • GPU Scale Up • Multi Node • Remote Spark • Batch Nodes • VM Scale Sets Scale Out
- 15. Series RAM vCPU GPU Approx Cost Standard_B1s 1 Gb 1 None Free [*] DS3_v2 14Gb 4 None $0.23 / hr DS4_v2 28Gb 8 None $0.46 / hr A8v2 16Gb 8 None $0.82 / hr Standard_NC6 56 Gb 6 0.5 NV Tesla K80 $0.93 / hr Standard_ND6s 112 Gb 6 1x Tesla P40 $2.14 / hr [*] Not recommended: Standard_B1s (free, but too small to be useful)
- 20. Azure Batch Batch pools Configure and create VMs to cater for any scale: tens to thousands. Automatically scale the number of VMs to maximize utilization. Choose the VM size most suited to your application. Batch jobs and tasks Task is a unit of execution; task = command line application Jobs created and tasks submitted to a pool; tasks are queued, then assigned to VMs. Any application, any execution time; run applications unchanged. Automatic detection and retry of frozen or failing tasks.
- 21. Cost savings Scale clusters size up and down as needed Reserved Instances for persistent infrastructure Per-second billing for VMs Flexible consumption and savings with low- priority VMs
- 22. Scaling AI with DSVM and Batch AI DSVM (Dev/Test Workstation) Azure File Store Azure Batch AI Cluster Batch AI Run Script Store Py Scripts in File Store Create Py Scripts Trained AI Model
- 28. BRK3320 The Developer Data Scientist – Creating New Analytics Driven Applications using Apache Spark with Azure Databricks May 8 10:30 AM-11:45 AM, Sheraton Grand Ballroom A
- 29. • Traditionally, static-sized clusters were the standard, so compute and storage had to be collocated • A single cluster with all necessary applications would be installed onto the cluster (typically managed by YARN, or something similar) • The cluster was either over-utilized (jobs had to be queued due to lack of capacity) OR was under-utilized (there were idle cores that burned costs) • Teams of data-scientists would have to submit jobs agaisnt a single cluster - this meant that the cluster had to be generic, preventing users from truly customizing their clusters specifically for their jobs Traditional / On-Premise Paradigm DataStore
- 30. • With cloud computing, customers are no longer limited to static size clusters • Each job, or set of jobs, can have its own cluster so that a customer is only charged for the minutes that the job runs for • Each user can have their own cluster, so that they don’t have to compute for resources • Each user can have their own custom cluster that is created specifically for their experience and their workload. Each user can install exactly the software they need without polluting other user’s experiences • IT admins don’t need to worry about running out of capacity or burning dollars on idle cores Modern / Cloud Paradigm DataStore
- 34. Connect to the Spark cluster: library(sparklyr) cluster_url <- paste0("spark://", system("hostname -i", intern = TRUE), ":7077") sc <- spark_connect(master = cluster_url) Load in some data: library(dplyr) flights_tbl <- copy_to(sc, nycflights13::flights, "flights") Munge with dplyr: delay <- flights_tbl %>% group_by(tailnum) %>% summarise(count = n(), dist = mean(distance), delay = mean(arr_delay)) %>% filter(count > 20, dist < 2000, !is.na(delay)) %>% collect
- 37. > m <- ml_linear_regression(delay ~ dist, data=delay_near) * No rows dropped by 'na.omit' call > summary(m) Call: ml_linear_regression(delay ~ dist, data = delay_near) Deviance Residuals:: Min 1Q Median 3Q Max -19.9499 -5.8752 -0.7035 5.1867 40.8973 Coefficients: Estimate Std. Error t value Pr(>|t|) (Intercept) 0.6904319 1.0199146 0.677 0.4986 dist 0.0195910 0.0019252 10.176 <2e-16 *** --- Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 R-Squared: 0.09619 Root Mean Squared Error: 8.075 >
- 38. cda.ms/sf
Editor's Notes
- #10: Can go download and use this today. All goodness of VSCode, etc. is part of this
- #12: Tools: - R/Python/Julia/etc. - Data platforms (SQL Server 2016 on Windows, Azure Data Lake and HDInsight tools in VS 2015, Spark and Hadoop on Linux), plus tools like Squirrel SQL, ODBC/JDBC drivers - Data movement tools: these, plus Azure CLI - ML + AI: deep learning + GPU support on new images. LightGBM. MicrosoftML. MRS. We enable many workflows. Also use it to experiment with new tools. Discuss: Windows 2012 versus 2016, Ubuntu vs CentOS. Deep learning + GPU support on new images. https://docs.microsoft.com/en-us/azure/machine-learning/machine-learning-data-science-virtual-machine-overview
- #14: Tools: - R/Python/Julia/etc. - Data platforms (SQL Server 2016 on Windows, Azure Data Lake and HDInsight tools in VS 2015, Spark and Hadoop on Linux), plus tools like Squirrel SQL, ODBC/JDBC drivers - Data movement tools: these, plus Azure CLI - ML + AI: deep learning + GPU support on new images. LightGBM. MicrosoftML. MRS. We enable many workflows. Also use it to experiment with new tools. Discuss: Windows 2012 versus 2016, Ubuntu vs CentOS. Deep learning + GPU support on new images. https://docs.microsoft.com/en-us/azure/machine-learning/machine-learning-data-science-virtual-machine-overview
- #15: Typical data scientist coding workflow looks something like this: You start small with a single DSVM. You perfect the code on just a subset of the data. Don’t worry about big data right away to keep good code pace. Once you are satisfied code works on a single machine, try to scale it up to larger VMs. Can GPUs or HPC type configurations help. And finally start working with your full dataset. You should have a good idea what kind of config you may need based on the single VM and scale up scenarios. From your DSVM desktop you can connect to remote Spark nodes, submit jobs to a batch pool, leverage a scale set which can autoscale. Bottom line is that ue DSVM for the tools, use different Azure services to help you scale up and scale out.
- #21: Azure Batch provides APIs for creating pools of resources, and then scheduling jobs and tasks to those resources. And the best part is that there is no charge for using Batch: you just pay for the compute and storage resources.
- #22: We understand that cost savings are paramount for customers. Azure offers flexible consumption. Mix and match low-priority VMs at discounted rates with on-demand VMs, along with per-minute billing to address with your priorities and budget. Our portal offers highly granular insight into your usage, associated costs, and groups using your resources. The Azure built-in, policy-based governance helps create a rich and integrated collaborative experience.
- #23: Talk about complete end to end development using DSVM. Left: Shared Data Stores, both cloud and on Prem. Center: DSVM as dev environments in the cloud Right: Trained Models and Code deployed from DSVM to Other Production systems or DSVMS used in Production as well.
- #29: R is #8 in January 2018 Tiobe language rankings. #6 in IEEE Spectrum 2017 top programming languages.