animind - Fotolia
The three major cloud services -- AWS, Google Cloud and Microsoft Azure -- are growing at double-digit rates as businesses speed up workload migrations, SaaS adoption and cloud-native application development. The increasing importance of online operations and digital systems to business success has broadened the use of cloud compute instances and storage.
Costs are a critical consideration -- often the deciding factor -- when weighing cloud storage services against traditional on-premises equipment. Unfortunately, evaluating cloud storage is more like shopping for a car than buying a new toaster. It's complicated, requiring analysis of myriad service options, pricing models and use cases.
Such complexity makes it impossible to provide unqualified recommendations given the diversity of offerings and uniqueness of every organization's needs. Instead, we look at AWS vs. Azure vs. Google pricing and storage features. Our goal is to outline the major storage product categories, pricing models and costs to help cloud storage buyers create a personalized ROI and TCO analysis and isolate the best service for their requirements.
The providers, service categories and cost models
AWS, Azure and Google Cloud dominate the market for cloud infrastructure services, garnering 60% of the worldwide revenue, according to Synergy Research Group. Their size and development budgets mean the three providers set the de facto standards for service offerings, features and pricing that other providers -- many of them technology giants in their own right -- emulate. Although our summary focuses on AWS vs. Azure vs. Google pricing and features, understand that other cloud service providers have similar products that are sometimes differentiated on price or niche features.
Storage services fall into three categories:
- Object storage is a system designed for unstructured data that treats it as granular information objects rather than as part of a file or volume. Objects are identified by a globally unique identifier, a collection of metadata and, in some implementations, a bucket that sets security and access policies for objects within it.
- File storage is analogous to the file systems used on PCs and servers that store data in files organized into hierarchical directories. Cloud file storage exposes files using standard NAS protocols, typically NFS and SMB. Some file storage implementations incorporate a managed storage server into the service to guarantee compatibility with Windows or Linux applications, while others use proprietary file system and protocol implementations.
- Block or volume storage is the cloud equivalent of a raw disk volume accessed via a SAN, although it's accessed via cloud APIs rather than iSCSI or Fibre Channel protocols. VM instances can attach to block resources to create database volumes, VM file systems or application-specific formats for media editing, geographic information systems, email repositories and VM image snapshots.
For object storage, AWS offers S3. It's the quintessential object service, with APIs that have become a de facto standard supported by third-party services and applications. S3 has multiple performance and price points, including standard, auto-tiering, infrequent access (both multiple and single availability zones) and Glacier for archiving.
Azure's object storage offering is Blob Storage. It's available in four performance tiers -- premium, hot, cool and archive -- each with pay-as-you-go or reserved capacity purchase plans. Volume discounts apply across all Azure Blob options.
Google Cloud object storage is available in four performance tiers: Standard, Nearline, Coldline and Archive.
AWS' file storage offering is Elastic File System, a managed, petabyte-scale NFS service with multi-availability zone redundancy. It's available in Standard and Infrequent Access modes that include an auto-tiering option. Azure's Files is like its Blob Storage in offering four performance tiers with added charges for optional snapshots and metadata backup. Google Cloud offers Filestore, a managed NFS service available in three performance tiers: HDD, SSD and High Scale SSD that provide a range of IOPS and maximum I/O throughput.
AWS also offers a file storage server service. Its FSx for Windows or Lustre are two variants of managed file servers, targeting apps that require tight integration with a storage server. FSx for Windows delivers SMB shares from a Windows server. FSx for Lustre provides a Linux-based Lustre file system for high-performance computing workloads. These services are available in HDD and SSD performance and single and multiple availability zone configurations.
For block storage, Amazon Elastic Block Store provides SSD- and HDD-based block devices at various IOPS and latency levels. Azure's block offerings include Managed Disks and Page Blobs. Managed Disks are designed for VM attachment and available in ultra, premium and standard SSD tiers, as well as an HDD tier. Page Blobs are a collection of 512-byte pages optimized for random read/write operations similar to a database and available in SSD and HDD tiers. Google offers Persistent Disk block storage in eight tiers that define performance and reliability parameters. Zonal persistent disks replicate data within a zone, while regional ones synchronously replicate data between two zones in the same region.
AWS vs. Azure vs. Google pricing comparison
Cloud service pricing is complicated. To provide a simple approach, we compared the monthly prices of basic configurations of each provider's storage product using cost calculators from AWS, Azure and Google Cloud (see table below).
These examples use basic configurations with as few optional or unique features as possible to simplify the analysis and make comparisons as fair as possible. For the configuration provided, Google Cloud is the least expensive for object and file storage services. Azure comes out as the least expensive for block storage. AWS is consistently in the middle.
Each service offers myriad configurations, options and price reduction options, such as prepaid or reserved capacity and lower-performance price tiers. Together, these provide enterprise customers tremendous flexibility in customizing storage performance, availability and redundancy to meet application and budget requirements.
As our examples demonstrate, the three major providers are competitive. When looking at AWS vs. Azure vs. Google pricing, significant cost differences are generally due to quirks in how the providers bundle storage features and price usage-based parameters, like read/write requests and snapshot storage. Storage buyers should always build detailed pricing models using the available online calculators before making cost-based buying decisions.