Useful range of products!
January 31, 2019
Useful range of products!
Score 9 out of 10
Vetted Review
Verified User
Overall Satisfaction with Microsoft Azure
We use select Microsoft Azure products and APIs to complete some of our more machine learning-oriented tasks off-premise. This includes modeling data we receive from magnetometer sensors, recognizing environmental effects in data during hardware testing, and some other backend-related tasks like modeling traffic and parking behavior. Using these products allows us to move tasks that require lots of processing power to cloud products that are optimized for this purpose.
- Scalable pricing -- The Azure pricing scales with usage, so the cost per month becomes very clear early on, and the ML-related products for Azure are quite competitively priced.
- Security management on Microsoft Azure is better compared to other cloud platforms I've seen, and it's really easy to configure.
- Startup Programs -- It's easy to get credits to try out Microsoft Azure through the Microsoft/Azure for Startups program. No other service provides this access and support for startups for free.
- Poor documentation -- Microsoft's documentation can take a while to get used to, as the format and tutorials are a bit different compared to AWS and other cloud computing platforms.
- Some code interfaces/SDKs are not well-designed. Specifically, the Python and Java SDKs can be quite difficult to integrate.
- It's difficult to onboard, due to the lack of beginner-oriented documentation for some of the ML products. Some of the products require extensive knowledge of how to use Azure in production.
- Positive -- Offloading specific ML-related tasks like Azure Machine Learning Studio was great for optimizing our general compute instances infrastructure.
- Negative -- Difficult to onboard and poor documentation, integrating some of the Azure products took significantly more time than necessary.
- Positive -- The pricing for some of the Azure products is quite competitive at greater usage.
Google Cloud has a selection of products that are specific to machine-learning, but they are better for more specific use-cases like Natural Language Processing, Computer Vision, and other APIs. For our company, we were looking for APIs and cloud products that had a bit more flexibility in their interface, which is something that Azure offered for more of their products.