We’re often asked, “What are the differences between AWS and Azure?” The question generally comes from clients looking to decide which path to head down, a decision that depends on figuring out the best-long term solution to support a particular workload and satisfy a client’s business needs and drivers.

At the end of the day, either AWS or Windows Azure could be a fit. Sometimes the answer comes down to feature/functionality needed, price or plain old preference. Other times it’s about terminology: Every cloud vendor has its own lingo, and this be can be a source of frustration when trying to compare services. If a client has taken the time to understand the language of AWS, it takes a bit of a time and effort to grasp the language of Azure.

Before we look at some areas of similarities and differences between the two innovative providers and their services, a little history.

Roots & Current Reality

In 2014, Gartner listed Amazon Web Services (AWS) and Microsoft as leaders in its Magic Quadrant for their ability to execute and completeness of vision. While Microsoft moved from visionary to leader in year-over-year growth, it’s second to AWS for four key reasons: cloud leadership, compute capacity, cost-effectiveness and pace of innovation.

Beyond terminology, there are fundamental differences between the two platforms. AWS offers 36 distinct, marketed capabilities across eight categories. AWS does offer a PaaS offering, Elastic Beanstalk, but the core competency is IaaS. Azure offers 68 distinct, marketed capabilities across seven categories. Of these, some are standalone. However, Microsoft owns the underlying software that these 68 capabilities are built on, so the integration has only one vision.

Service Models

Let’s take a minute and review a fundamental difference. There are three primary cloud service models: infrastructure as a service (IaaS), platform as a service (PaaS) and software as a service (SaaS). AWS’ roots are primarily IaaS; that’s because of its storage footprint and the demand of supporting Simple Storage Service, Linux, Firefox and SimpleDB.

In contrast, Azure’s focus was primarily PaaS due to the launch of SQL 2008, SharePoint and .Net integration points. As a rule of thumb, in IaaS, the cloud provider deals with Layer 1 through Layer 4 of the OSI model, and the client controls Layers 5 through 7. With PaaS, the cloud provider handles Layers 1 to 5, and the customer controls Layers 6 and 7. That means that an AWS customer controls the OS, data and applications, while an Azure customer controls applications and data.

The difference comes down to the abstraction layer: PaaS handles the infrastructure and the runtime environments. This abstraction from the physical environment gives PaaS the advantage in terms of scalability, with less overhead management. A client manages the application and its data, and the rest is handled by the provider.

There are some other key differences: AWS is VM-first while Azure is services-first. Azure has a clear advantage for customers looking for PaaS. As for mobility, Azure Mobile Services are mature and full-featured. Azure brings native API management and full application integration, while AWS delivers native OLAP data warehousing.

Service Categories

AWS and Azure divide services into four main categories: compute, storage, database and networking. Additionally, there are deployment and configuration management services (Opworks, CloudFormation), CDN (CloudFront) and application development platforms. Moving down the stack, things can get foggy.

Compute

AWS has EC2, while its closest neighbor at Azure is VM Role. This is local instance store. Both are stateless instances that don’t persist data on the root disk. Both support a persistent file model. Web and Worker rules are pure PaaS instances. The client publishes an application to them, and then the underlying platform is maintained by the client.

Web Roles in PaaS are best used for front-end Web applications. A client can quickly and easily deploy to Web Roles.

Worker roles, meanwhile, are best suited for perpetual or long-running tasks. These tasks run asynchronously in the background. The goal with this separation is to maximize agility and control. The thinking is that when the application’s background process in a Worker role is separated from the front end in a Web Role, application logic is more evenly distributed. This allows for better control when scaling the application.

Simply put, AWS has the ability to support grid computing that runs multi-node tasks and tends to support longer processes. Azure is built around services running natively on the OS for resilience and uses multitasking nodes with shorter processes. Again, for Amazon, the focus was support for Linux and large-batch processing, while Azure support was built on Microsoft’s own OS and application stack.

Storage

Microsoft 365 has OneDrive and Azure Storage, while AWS has EBS. Both align in a similar fashion and are block storage vehicles.

AWS’ Simple Storage Service (S3) is comparable to Azure Blob. Both serve the same purpose: Store large, binary objects. Clients sometimes refer to this as “archival storage.” AWS also offers Glacier, a tape-to-disk cold storage offering intended for very large data sets.

Database

AWS Dynamo is a NoSQL database that is similar to Azure Table Storage. NoSQL is a great choice when storing unstructured data in a cloud environment. Both providers’ offerings scale well.

Relational Database Service (RDS) is a traditional relational database in the cloud offered by AWS. When a customer needs to store structured data with transactional support, this is the service for them. Azure’s answer to RDS is SQL Azure; SQL has been a staple in most business and is proprietary.

Other AWS database products are Redshift and SimpleDB.

Queue Services

Azure has Azure Queue, while AWS provides SQS. Queuing is vital to a successful cloud scale-out. It is the primary way your machines relay information to one another, so don’t skip evaluating this area.

In the end, both suppliers offer purpose-built clouds that have evolved and grown. Their core cloud automation and support choices, however, stay true to the companies’ respective roots. Additionally, both infrastructures are running on their own technologies and don’t need to deal with the decisions and direction of underlying virtualization and tooling. They are in control of their own futures, and that reflects in their top-tier placement in Gartner Magic Quadrant and other sources.

Michael Goodenough is vice president of cloud solutions at BCM One. He is a is a cloud visionary who combines years of technological experience with a deep knowledge of the global marketplace and a clear vision of the future.

Jo Peterson is the vice president of converged cloud and data for Clarify360, a division of Teleproviders, a southern California master agent. She joined the company in 2009 as a regional sales manager with responsibility for broadening the carrier portfolio, developing strategic vertical sales efforts and aligning a go-to-market strategy in the cloud services space.