Companies continue to make the move into cloud computing. Whether it’s using an individual cloud service or migrating your entire infrastructure into a new cloud ecosystem, you’re not alone in seeking out the added benefits of cloud technologies.

From improvements in scalability, security, and flexibility to reductions in cost and environmental impact, there are a wealth of reasons to make the move into the cloud. Of course, the transition is not as easy as it once was.

Since its inception, the cloud ecosystem has become a complex, ever-expanding myriad of providers, technologies, products, and services. As you attempt to piece together the different combinations across these verticals your choice of options can quickly climb into the 1000s. It quickly becomes apparent there is a thing as too much choice.

Cloud Computing Companies and Services (Image source: imelgrat.me)

Like in any industry, a handful of companies rise above the rest to become market leaders. When we think about cloud computing providers, there are three names that top the list: Google Cloud Platform, Amazon Web Services, and Microsoft Azure.

Today, we will be comparing two cloud giants, Google Cloud Platform and Amazon Web Services. We’ll be taking a deep dive into the products and services of each provider. Seeking to add clarity and simplify the process comparing these two cloud providers in order to make an informed decision.

Although we exclusively utilize the Google Cloud Platform here at Kinsta, we’ll be providing you with an unbiased opinion. Both platforms offer extensive benefits, but which is right for you will ultimately depend on your company’s own unique requirements.

Why Google Cloud vs Amazon Web Services

If you’re planning on using cloud services, the three providers you will undoubtedly discover Google Cloud, Amazon Web Services, and Microsoft Azure. Today, we will be focussing on comparing two of these, specifically Google Cloud vs AWS.

These cloud giants are household names in the tech space. Both organizations have dominated for more than a decade in their respective industries. Renowned as world-leading companies, they are meticulous in their pursuit of innovation and excellence. Each boasting a wealth of tech industry expertise that is near impossible to compete with.

With their respective technological foundations, it is unsurprising they have developed industry-leading cloud computing platforms. In September 2020, Gartner has again named Google and AWS as leaders in their Infrastructure as a Service (IaaS) Magic Quadrant.

For Amazon, this is the 10th consecutive year AWS has secured the top-right corner of the Leader’s quadrant in Gartner’s Magic Quadrant for Cloud Infrastructure as a Service (IaaS). Earning the highest placement for Ability to Execute and furthest for Completeness of Vision.

2020 Magic Quadrant for Cloud Infrastructure as a Service
2020 Magic Quadrant for Cloud Infrastructure as a Service, Worldwide (Image source: Gartner)

Google Cloud and AWS Continue to Dominate the Industry

Google Cloud and AWS have dominated the cloud computing space since IaaS solutions began to gain traction in 2008.

In August 2020, a report from Gartner named both Google and Amazon in a group of 5 public cloud infrastructure providers that make up 80% of the IaaS market. A trend that is only set to continue as both organizations double down to consolidate their foothold in the market.

Despite the global pandemic stalling major economies, Gartner is forecasting worldwide public cloud revenue growth in 2020 by 6.3%. Driven by an explosion in remote working, we can reasonably expect comparative results in the cloud space. Especially with the report outlining a 94% increase in the Desktop as a Service (DaaS) market. Against this backdrop, you can expect Google and Amazon to continue expanding.

While they both started life in the IaaS space, you can now turn to Google Cloud and AWS for 100s of solutions across IaaS, SaaS, and PaaS. With both organizations continuing to innovate and add new cloud services offerings to their ever-expanding roster.

Google Cloud Platform Revenue

Alphabet’s Q4 and Fiscal Year 2019 results showed the company continued to deliver strong growth, with overall revenue up 18% year-on-year. While there is a lack of transparency in revenues contributed by Google Cloud, the company reported impressive growth in excess of 100%, putting the company on an annual run rate of $10 billion as of year-end.

In 2020, the arrival of the Coronavirus pandemic saw Google Cloud’s parent company – Alphabet – record its first-ever quarterly revenue decline since going public in 2004. Against this dire backdrop, Google Cloud has in fact bucked the trend, seeming only to have accelerated growth.

In Q1, Google Cloud made significant gains thanks to Google Meet, when their video conferencing tool became a hit for remote workers. Earning release statements for Q1, Q2, and Q3 show a pattern of continued YoY revenue growth for the Google Cloud Platform. As we head towards the end of 2020, Google Cloud revenue is projected to grow to an annual run rate in excess of $13 billion – a predicted 30% growth in 2019.

Amazon Web Services Revenue

In 2019, Amazon’s Q4 Earnings Release reported AWS sales revenue of nearly $10 billion. Putting the organization on an annual revenue run rate in excess of $40 billion.

With the emergence of the Coronavirus pandemic in 2020, AWS growth has slowed significantly. With Q1, Q2, and Q3 earnings release statements showing YoY growth declining and settling into a sub 30% growth rate in each respective quarter. This is a marked slow down from growth of 40-50% over the previous 3 years.

This is hardly a doom and gloom scenario, AWS is now on a $43 billion annual revenue run rate with the figure expected to expand once Q4 is complete. The exception might be if you’re an Amazon shareholder, especially after Jeff Bezos told Amazon shareowners to “take a seat” while their COVID-19 response eats into operating profits.

Google Cloud vs Amazon Web Services Features Comparison

It is no simple task to compare the Google Cloud vs AWS platforms. Their sprawling and ever-expanding cloud services now include 100s of products from which to choose from. Complicating matters further, the providers often use different naming conventions for comparative products. So, to avoid getting lost in the detail, it requires a certain level of knowledge and understanding.

Simplifying the task, both the Google Cloud Platform and Amazon Web Services Platform thankfully group their products under the same category headings. Accelerating the process to save you time, we’ve done the heavy lifting of comparing the most commonly used services from business-critical categories.

In this section, we’ll explore the products that combine to create a typical cloud deployment – compute, networking, security, and storage. Here at Kinsta, we have the first-hand experience of utilizing these services in delivering market-leading hosting solutions.

We also cover the critical considerations that surround these services. Service support, platform stability, pricing, and billing structure.

Compute Features

When comparing Google Cloud vs Amazon Web Services compute capabilities, we’ll be focussing on virtual machines (VMs).

These computer system emulations provide the functionality of a physical computer and run almost any workload you can think of. They are the foundation of your cloud environment, it’s critical you choose a VM setup that suits your business needs.

Both cloud providers have adopted a similar approach to VMs, though they use different naming conventions for their individual product offerings.

Compute Engine is the service offering on the Google Cloud Platform, while Amazon Web Services is named Amazon Elastic Compute Cloud (Amazon EC2). Each provider also uses different terminology and concepts.

Thankfully, Google has mapped Amazon EC2’s terminology and concepts to that of Compute Engine – which you can see in the table below:

Feature Amazon EC2 Compute Engine
Virtual machines Instances Instances
Machine Images Amazon Machine Image Image
Temporary virtual machines Spot instances Preemptible VMs
Firewall Security groups Compute Engine firewall rules
Automatic instance scaling Auto Scaling Compute Engine autoscaler
Local attached disk Ephemeral disk Local SSD
VM import Supported formats: RAW, OVA, VMDK, and VHD Supported formats: RAW, OVA, VMDK, and VHD
Deployment locality Zonal Zonal

Mapping high-level terminology for Amazon EC2 to Google Compute Engine (Table Source: Google)

Virtual Machine Features

When deploying virtual machine instances on Compute Engine Amazon EC2, both services deliver many features that closely align, these include:

  • The ability to use stored disk images to create instances
  • On-demand capabilities to launch and terminate instances
  • Restriction free management of your instances
  • The ability to tag your instances
  • A variety of available operating systems that can be installed on your instance

Virtual Machine Access

When it comes to accessing your VM, there are a number of key differences in the approach taken between Compute Engine and Amazon EC2.

If you want terminal access to an instance in Amazon EC2 you’ll need to include your own SSH key.

Compute Engine offers a more flexible approach to terminal access. Allowing you to create an SSH key as and when you need it, even if that instance already running. You also won’t need to store these keys on your local machine, thanks to Compute Engine’s browser-based SSH terminal which is available via the Google Cloud Console.

Virtual Machine Instance Types

When deploying your virtual machine, both Compute Engine and Amazon EC2 offer simplicity through a range of predefined instances. These instances incorporate specific configurations of virtual CPU, RAM, and network.

Both Google and Amazon offer 100s of virtual machine types available in a variety of configurations. Each offers flexibility, allowing you to customize your configurations in order to scale your VM resources to meet the unique needs of your business.

You can do this by increasing the number of CPUs and available RAM to extreme high-end specifications.

The providers max out with the following:

  • Google Compute Engine VMs scaling up to 416 vCPUs and 11,776 GB of RAM
  • Amazon EC2 VMs scaling up to 448 vCPUs and 24,576 GB of RAM

Across the range of VM types, both platforms use largely the same categorization. Though in certain categories, one provider may offer a machine type the other does not.

Depending on your business requirements, you can choose from machine types across categories including shared core, general-purpose, memory-optimized, compute-optimized, storage optimized, GPU, and high-performance categories.

To provide you with the best VM comparison between Amazon EC2 and Compute Engine, we’ve compiled the following table which lists the most up-to-date machine types for both services.

Machine Type Amazon EC2 Compute Engine
Shared Core N/A f1-micro – g1-small
e2-micro – e2-medium
General Purpose a1.medium – a1.metal
t4g.nano – t4g.2xlarge
t3.nano – t3.2xlarge
t3a.nano – t3a.2xlarge
t2.nano – t2.2xlarge
m6g.medium – m6gd.metal
m5.large – m5d.metal
m5a.large – m5ad.24xlarge
m5n.large – m5dn.24xlarge
m4.large – m4.16xlarge
e2-standard-2 – e2-standard-32
e2-highmem-2 – e2-highmem-16
e2-highcpu-2 – e1-highcpu-32
n1-standard-1 – n1-standard-96
n1-highmem-2 – n1-highmem-96
n1-highcpu-2 – n1-highcpu-96
n2-standard-2 – n2-standard-80
n2-highmem-2 – n2-highmem-80
n2-highcpu-2 – n2-highcpu-80
n2d-standard-2 – n2d-standard-224
n2d-highmem-2 – n2d-highmem-96
n2d-highcpu-2 – n2d-highcpu-224
Memory-optimized r6g.medium – r6gd.metal
r5.large – r5d.metal
r5a.large – r5ad.24xlarge
r5n.large – r5dn.24xlarge
r4.large – r4.16xlarge
x1e.xlarge – x1e.32xlarge
x1.16xlarge – x1.32xlarge
u-6tb1.metal -u24tb1.metal
z1s.large – z1d.metal
m1-ultramem-40 – m1-ultramem-160
m1-megamem-96
m2-ultramem-208 – m2-ultramem-416
Compute-optimized c6g.medium – c6gd.metal
c5.large – c5d.metal
c5a.large – c5ad.24xlarge
c5n.large – c5n.metal
c4.large – c4.8xlarge
c2-standard-4 – c2-standard-60
Storage-optimized i3.large – i3.metal
i3en.large – i3en.metal
d2.xlarge – d2.8xlarge
h1.2xlarge – h1.16xlarge
N/A
GPU p4d.24xlarge
p3.2xlarge – p3db.24xlarge
p2.xlarge – p2.16xlarge
inf1.xlarge – inf1.24xlarge
g4dn.xlarge – g4dn.metal
g3s.xlarge – g3.16xlarge
f1.2xlarge – f1.16xlarge
NVIDIA® Tesla® T4 – NVIDIA® Tesla® K80
NVIDIA® Tesla® T4 Virtual Workstation – NVIDIA® Tesla® P100 Virtual Workstation
High performance N/A N/A
Custom VM resource configuration Yes Yes

Virtual Machine Images

To accelerate your virtual machine deployment you can use machine images.

These are typically configured to include an operating system and the required supporting web server and database software. Both Compute Engine and Amazon EC2 use machine images to create new instances. In addition to the standard configurations, they both allow you to use images published by a third-party vendor or custom images created for private use.

The platforms are similar enough that you can use the same workflow for image creation on both Amazon EC2 and Compute Engine.

When it comes to image storage, they take slightly different approaches. On Google Cloud, images are stored with Compute Engine, while Amazon EC2 stores its images in different services – Amazon Simple Storage Service (S3) or Amazon Elastic Block Store (EBS).

The distinct benefit Amazon EC2 offers over Compute engine is the ability to access a community repository of ready-made images and the ability to make your own images publicly available (should this be a requirement).

On the flip-side, Compute Engine offers the benefit of globally available machine images. While Amazon Machine images are geo-locked, meaning they are only available in a specific region.

Automatic Instance Scaling of Virtual Machines

One of the most powerful cloud benefits is the ability to scale your workload resources to meet demand. This goes both ways, increasing resources in peak periods to maintain performance and inversely reducing resources in quiet times to limit wastage and control spend. This process is widely known as autoscaling.

Both Compute Engine and Amazon EC2 support and implement autoscaling similarly, allowing you to create and remove resources in line with user-defined policies.

Amazon EC2 auto-scales instances in a group, with each instance created from a defined launch configuration. Instances are created or removed based on one of three chosen scaling plans

  • Manual – you manually instruct auto-scaling up or down
  • Schedule – you configure specific timeframes to auto-scale resources
  • Dynamic – you create policies to scale your instances based on Amazon CloudWatch metrics or Amazon Simple Queue Service (SQS) queues.

Compute Engine scales instances in a managed instance group. Each instance group is created from an instance template with resources scaled based on an autoscaling policy. Unlike Amazon EC2, Compute Engine’s auto scaler only supports dynamic scaling.

Temporary Virtual Machine Instances

If you want to tap the power of cloud computing, but only have a limited budget, it’s worth exploring the option of temporary instances. Virtual machines that are running on spare cycles of resources allocated to other processes.

Temporary instances are available sporadically, so are best used on jobs that:

  • can be interrupted without you losing work
  • don’t need completing in a set timeframe, typically low priority workloads
  • don’t need higher computational power, such as rendering video

Both Amazon EC2 and Compute engine offer a version of temporary instances. Though they use different pricing models and naming conventions, they share a set of common attributes when their temporary VMs:

  • are fully controllable while running
  • run at the same performance levels as on-demand instances
  • are restricted to a subset of machine types and machine images versus on-demand instances

Amazon EC2 temporary VMs are known as Spot Instances. They are available in two formats:

  • Undefined Spot Instances – you purchase a Spot Instance for an undefined period of time, paying the price that is in effect for the period your instances are running. This type of instance can be available at a discounted price of up to 90% of standard on-demand pricing. You can check and compare current Spot prices versus On-Demand rates via the Spot Instance Advisor.
  • Spot Instances for the predefined duration—you purchase a block of time in advance. Available in hourly increments for up to 6 hours. With forward planning, you only access discounts ranging from 30-50%.

Compute Engine temporary VMs are named Preemptible Virtual Machines. They are available longer than their Amazon EC2 counterparts, running for up to 24 hours (if not reclaimed) before being automatically terminated. Their pricing structure is fixed and is available at a discounted rate of up to 80% versus the on-demand rates of equivalent VM instances.

Networking Features

Amazon Web Services and Google Cloud have each developed a formidable global cloud infrastructure. Their sprawling networks consist of 100s of interconnected data centers across the globe.

Each provider has developed a state-of-the-art cloud network designed for high fault tolerance, countless redundancy scenarios, and low latency levels. Each offers networking services capable of delivering high-speed connectivity to VMs, other cloud services, and on-premises servers.

Within this section we will take a closer look, comparing the networking products and services on offer from Google and Amazon.

Product Amazon Web Services Google Cloud Platform
CDN Amazon CloudFront Cloud CDN
Dedicated Interconnection AWS Direct Connect Cloud Interconnect
DNS AWS Route 53 Cloud DNS
Load Balancing Elastic Load Balancing Cloud Load Balancing
Virtual Networks Amazon Virtual Private Cloud Google Virtual Private Cloud
Tiers N/A Network Service Tiers

Locations

Both providers continue a rapid expansion of their respective infrastructure, with new data center locations in development or planned for the future. When comparing the location numbers for network availability, it appears too close to call.

Google Cloud Network Locations

Google boasts cloud network locations currently available across 37 regions, 73 zones, 144 network edge locations, and 200+ countries and territories. They recently added new locations in Seoul, Salt Lake City, Las Vegas, and Jakarta.

The future will see Google Cloud continue to expand into the following locations: Warsaw (Poland), Doha (Qatar), Toronto (Canada), Paris (France), Milan (Italy), Santiago (Chile), and Madrid (Spain).

Google Cloud Regional Network
Google Cloud Regional Network (Image Source: Google Cloud)

Amazon Web Services Network Locations

AWS now offers cloud network locations available in 24 regions, 77 zones, 210 network edge locations, and 245 countries and territories. While the figures appear too close to call, Amazon’s network is bigger, offering multiple availability zones in twice as many regions as Google. Which would give them an edge when it comes to latency.

Coming soon, Amazon plans to launch additional data centers in Hyderabad (India), Jakarta (Indonesia), Osaka (Japan), Madrid (Spain), and Zurich (Switzerland).

AWS Regional Cloud Network
AWS Regional Cloud Network (Image Source: AWS)

 

Content Delivery Network (CDN)

AWS and Google Cloud each offer a Content Delivery Network (CDN) product. Both unlock the ability to deliver your content and services to end-users faster, by replicating and hosting it across their global infrastructure to allow for more localized access. This means quicker load times, reduced strain on bandwidth, and greater responsiveness across your applications, websites, and services.

Named Amazon CloudFront and Cloud CDN, they each offer enhanced security to defend against the most frequently occurring network and transport layer DDoS attacks by default. They also offer deep integration with their respective platforms allowing you to unlock additional tools to monitor and improve performance.

Load Balancing

Both Google Cloud and AWS offer load balancing services. Configured appropriately, they will help you to automatically distribute traffic across multiple instances for improved availability and fault tolerance of your applications. They offer these services in differing configurations which we will now look at more closely.

AWS Load Balancing

The load balancing service from AWS is called Elastic Load Balancing (ELB). It has the following characteristics and capabilities:

  • You can use AWS load balancing services both internally and externally.
  • It lets you direct traffic to instances in one or several availability zones in a specified region.
  • Regular health checks are carried out on target instances, when an instance becomes unhealthy, traffic is redirected.
  • ELB can be integrated with the AWS Auto Scaling Service, this allows the automatic addition and removal of the instance when Auto Scaling scales the up or down
  • An Application Load Balancer is available for content-based routing and SSL
  • A Network Load Balancer is available for high throughput, low latency, Layer 4 connections.

Check out the Elastic Load Balancing comparison section for a more detailed feature comparison.

Google Cloud Load Balancing

Googles’ load balancing service is aptly named Cloud Load Balancing. It offers differing characteristics and capabilities:

  • Google Cloud load balancing services are separated between internal and external access.
  • Unlike ELB, you’re given a single IP address, accessible globally, when any external Compute Engine load balancer is provisioned. This IP address is used for the lifetime of the load balance and so can be used for DNS records, allowlists, and configuration in apps.

The different types of Compute Engine of load balancers include:

  • Network load balancing – designed for external Layer 4 load balancing, it supports UDP and TCP traffic balancing across multiple ports or port ranges.
  • HTTP(S) load balancing with TCP and SSL proxy – designed for external Layer 7 load balancing, traffic is balanced through various global and regional protocols. With traffic automatically redirected to the nearest backend, based on available capacity.
  • Internal TCP/UDP load balancing – software-defined regional load balancing which redirects traffic from your instance to a backend instance.
  • Internal HTTP(S) load balancing – delivering proxy-based load balancing of Layer 7 application data, with advanced traffic management and TLS termination.

Private Connectivity to Other Networks

If you want to create a private connection to instances outside your cloud setup cloud, like your on-premises environment, both AWS and Google Cloud offer services for multiple requirements:

Virtual Private Network (VPN)

The respective offerings of Cloud Router and Amazon VPC allow you to create a private gateway between their cloud and your networks

Private Connectivity to a VPC

When a VPN doesn’t provide the speed you need for certain workloads, a dedicated resource is required. Both providers offer private connectivity services with a network line offering a dedicated capacity level:

  • AWS lets you create a privately leased line with an AWS partner through its Direct Connect service. Access a 1-10 Gbps connection offering you connection speeds from 50 Mbps.
  • Google lets you create direct physical connectivity to your Google VPC from a partner facility with 10 Gbps increments through its Dedicated Interconnect service. Like AWS, Partner Interconnect delivers connection speeds from 50 Mbps.

High-Speed Connectivity to Other Cloud Services

Both providers offer high-speed connectivity for access to cloud services outside your VPC.

AWS’s Direct Connect service creates a separate virtual interface through which you can access all AWS cloud services.

Google Cloud has a wider range of services:

  • Direct Peering – lets you access all Google cloud services through a private network line to any of Google’s Edge Points of Presence.
  • Carrier Peering – offers the same interconnectivity services, only the connection is leased from a Google Partner.
  • Private Google Access for on-premises hosts – delivers private access through Dedicated Interconnect or Partner Interconnect.

Content Delivery Network Connectivity

Both providers offer discounted egress rates from your cloud resources to a CDN provider. Amazon provides these rates for its own CDN service only, Amazon CloudFront. Google offers CDN Interconnect, which provides discounted egress rates through several CDN providers.

DNS

Both providers deliver managed DNS services through their respective Amazon Route 53 and Cloud DNS offerings. Each supporting nearly all DNS record types, anycast-based serving, and domain name registration.

Where they differ, Amazon Route 53 supports two routing options, where Cloud DNS does not. Geography-based routing, allowing you to restrict content to geographic locations. And latency-based routing, which directs traffic based on latency levels measured by the DNS services.

The table below outlines a list of features mapped across both services:

Feature Amazon Route 53 Cloud DNS
Zone Hosted Zone Managed Zone
Support for most DNS record types Yes Yes
Any-cast-based serving Yes Yes
Latency-based routing Yes No
Geography-based routing Yes No
DNSSEC for DNS Service No Yes
Private Zones / Split Horizon Yes Yes

Network Service Tiers

To date, the Google Cloud Platform is the only provider to offer network service tiers to its customers. Selecting between a Standard and Premium tier, you have the flexibility to optimize your network based on performance and price.

Premium Tier

Choosing the Premium tier unlocks Google’s high performance and low latency network. Your traffic is prioritized, being routed through the fewest hops via the fastest paths to accelerate transport speeds and increase security. You also gain access to global network load balancing, while being protected by a Global SLA.

Google-Cloud-Network-Premium-Tier
Google Cloud Platform Premium Tier (Image Source: Google Cloud)

Standard Tier

Choosing the Standard tier connects you to Google’s lower performance network, still highly competitive with other public cloud services. Your load balancing services remain regional and you are not protected by a Global SLA. This option is for those where cost outweighs performance considerations.

Google-Cloud-Network-Standard-Tier
Google Cloud Platform Standard Tier (Image Source: Google Cloud)

Storage Features

There are five different types of storage services available from the Amazon and Google Cloud platforms. Understanding the different storage and disk types utilized is important, as they will have a direct influence on your performance.

Distributed Object Storage

Distributed object storage is a method of storing data as objects, also known as blobs. It allows you to store, protect, and access large volumes of data for use across a wide range of scenarios including websites, mobile apps, backups, archiving, and big data analytics.

Amazon Simple Storage Service (S3) and Google Cloud Storage are the competing distributed object storage services. They each function similarly, allowing you to store objects in a bucket. Each bucket can be identified with a unique key, and each object has an associated metadata record containing information including object size, date of last modification, and media type.

Both providers also a similar feature set for their services including:

  • The ability to host static media and web content
  • Object Versioning – where an object can be stored as multiple distinct versions to prevent data loss through objects being accidentally overwritten
  • Object Lifecycle management – allowing you to automate the migration and deletion of objects through preset user-specified lifecycle policies
  • Update Notifications – which can be configured to issue notifications whenever objects are created, updated, or deleted. Google Cloud Storage offers a more granular approach to notification types.
  • Service Level Agreement (SLA) – both Amazon S3 and Cloud Storage provide SLA uptime guarantees with a tiered refund amount once uptime drops below 99.95%.

Below is a table outlining a more detailed comparison of terminology and features:

Feature Amazon S3 Cloud Storage
Unit of Deployment Bucket Bucket
Deployment identifier Globally unique key Globally unique key
File system emulation Limited Limited
Object metadata Yes Yes
Object versioning Yes Yes
Object lifecycle management Yes Yes
Update notifications Event notifications Pub/Sub Notifications for Cloud Storage, Cloud Storage triggers for Cloud Functions, and object change notifications
Service classes Standard, Standard-Infrequent Access, One Zone-Infrequent Access, Amazon Glacier Standard, Nearline, Coldline, Archive
Deployment locality Regional Multi-regional and regional
Pricing Priced by the amount of data stored per month, network egress, and number of common API requests Priced by the amount of data stored per month, network egress, and number of common API requests

Block Storage

Block storage is the process of adding a virtual disk to a cloud-based virtual machine.

Both providers offer block storage services that integrate with their respective VM compute services, offering multiple block storage types that can be configured to varying performance and pricing levels.

Google provides Persistent Disk in combination with Compute Engine for its block storage service. While Amazon provides Elastic Block Store (EBS) in conjunction with Amazon EC2. Each offers you the ability to attach disks in two different ways:

Network-Attached Disks

A network-attached disk is where a disk volume is connected to your VM instance via the cloud provider’s network. This brings the inherent cloud benefits of built-in redundancy, snapshotting, and ease of detachment and reattachment of disk volumes.

Let’s take a high-level look at the feature comparisons between Google and Amazon block storage services:

Feature Amazon EBS Google Persistent Disks
Volume types EBS Provisioned IOPS SSD, EBS General Purpose SSD, Throughput Optimized HDD, Cold HDD Zonal standard persistent disks (HDD), regional persistent disks, zonal SSD persistent disks, regional SSD persistent disks
Volume locality rules Must be in same zone as instance to which it is attached Must be in same zone as instance to which it is attached
Volume attachment A single volume to be attached to up to 16 instances – each with read-write permissions to the shared volume A single volume can be attached to up to 10 instances in read-only mode
Attached volumes per instance Up to 40 Up to 128
Maximum volume size 16 TiB 64 TB
Redundancy Zonal Zonal or multi-zonal depending on volume type
Snapshotting Yes Yes
Snapshot locality Regional Global

There are some distinct feature differences that require a closer inspection:

Volume Attachment and Detachment
Once you create a disk volume, you can attach it to a single Compute Engine or Amazon EC2 instance. This instance can then mount and format the disk volume. You can also choose to unmount and detach this disk volume, which can then be reattached to a separate instance.

Until recently, Google had a significant edge offering the capacity for a single volume to be attached to multiple instances in read-only mode. This changed with Amazon’s introduction of EBS Multi-Attach, allowing a single volume to be attached to up to 16 AWS Nitro-based instances inside the same availability zone. With each instance having read-write permissions to the shared volume.

Volume Snapshotting
Google Persistent Disk and Amazon EBS allow you to create and store snapshots of your disk volume. Allowing you to create new volumes at a later date using the snapshot.
The process of creating snapshots is similar across the services. Initially creating a full copy of the volume, with future snapshots only copying changes from the previous volume.

It’s their availability that differs. Google Snapshots benefit from being available globally and can be used in any region without additional charges or requirements. Amazon EBS snapshots are different, only available in one region by default. You must expressly copy and incur data transfer charges if you wish to make a snapshot with AWS available in another region.

Locally Attached Disks

A locally attached disk is directly connected to the physical machine running your instance. This direct connection offers the benefits of reduced latency and higher throughput of increased performance.

Let’s take a closer look at how locally-attach disks in Compute Engine and Amazon EC2 compare for features in their respective block storage services:

Block Storage Amazon EC2 Google Persistent Disks
Service name Instance store Local SSD
Volume attachment Tied to instance type Can be attached to any non-shared-core instance
Device type Varies by instance type – HDD, SSD, etc. SSD
Attached volumes per instance Varies by instance type – up to 24 24
Storage capacity Varies by instance type – up to 2500 GB per volume 356 GB per volume
Live migration No Yes
Redundancy None None

File Storage

If you need file storage as part of your cloud setup, there are services on offer from both AWS and Google Cloud. They’re respectively named Amazon Elastic File System (EFS) and Google Filestore, the latter a new edition emerging from beta testing in late 2018.

Both offer a fully managed service, where you can quickly create and configure file systems, while the underlying infrastructure and associated deployment, patching, and maintenance are handled by your provider. Security of data is ensured through encryption at rest and in transit, with the capability to scale your instances to meet changes in performance requirements.

A big difference between the two providers, Amazon EFS runs on the newer Network File System Protocol, NFSv4. While Google’s Filestore service utilizes the older NFSv3 protocol. While studies have shown NFSv4 provides significant performance enhancements over NFSv3, the throughput and IOPS performance specs from both providers are strikingly similar.

Amazon EFS offers performance capabilities up to 10 GB/sec and over 500,000 IOPS, while Filestore is said to max out at 16 GB/sec and 480,000 IOPS. In layman’s terms, both will offer you a storage infrastructure capable of handling your highest performance workloads with low latency.

Cool Storage

If you’re planning on storing data that is accessed infrequently, without the need for immediate availability, you should consider cool storage.

Both Amazon S3 and Cloud Storage provide a reduced-cost storage class for this type of data. Amazon S3 offers infrequent storage classes Standard-IA and One Zone-IA. While Cloud Storage offers infrequent classes of Nearline and Coldline.

Cold or Archival Data Storage

If you’re planning on storing data for archival purposes, that doesn’t require regular or quick retrieval, both Amazon and Google offer an additional cold storage class for this data type. They are known as Amazon Glacier and Google Archival Cloud Storage.

Both are cost-effective storage options for the long-term preservation of data that is perhaps accessed less than once a year.

Security Features

Security will be chief among your considerations when exploring a cloud provider. When comparing cloud security, you want to explore and focus on the controls, policies, processes, and technologies that will combine to protect your cloud-based data, systems, and infrastructure.

Both Amazon Web Services and Google Cloud are renowned for offering cutting-edge cloud security. Committed to continually advancing research and development of their platforms to remain resistant to an ever-evolving threat landscape.

Taking a high-level approach, both providers deliver cloud security in three ways:

  • Security of their cloud – delivering you protection by default through security capabilities built into the underlying infrastructure of their cloud platform
  • Security in the cloud – allowing you to enhance the protection of your applications and data through additional security products and services available within their cloud platform
  • Security anywhere – protecting your assets regardless of location, by expanding security capabilities beyond their cloud platform with protocols like encryption

Delving beneath the service, let’s explore the main considerations when comparing Google Cloud Security vs AWS Security.

Compliance

Data protection and compliance are an ever-rising tide of regulatory control applied to information by governments and industry alike. Compliance has to be considered when choosing your cloud platform.

Both the AWS and Google Cloud platforms meet some of the toughest compliance requirements including CSA STAR, GDPR, HIPPA, PCI-DSS, and a range of ISO standards.

Both providers offer compliance programs that encompass certifications, law, regulations, frameworks, and privacy with a distinct crossover.

Amazon’s cloud platform and AWS compliance programs meet 75 compliance standards. Google Cloud’s compliance offerings also deliver against 75 compliance standards. Making both providers a viable option, even if you’re working in a heavily regulated sector like healthcare or financial services.

Continuing to steal headlines in the realm of compliance is GDPR. Rest assured both AWS and Google Cloud platforms are GDPR compliant, each offering a resource center. Helping you meet the protection, processing, and privacy requirements of any data you hold of a European citizen.

Encryption

Encryption plays a critical role in protecting your information. The practice of encoding data – making it near impossible to decipher without a decryption key – should be implemented regardless of where your data is held. Ensuring your data is safe even if it’s intercepted in-transit or at rest

Both Google Cloud and AWS offer encryption by default for data-in-transit and at-rest using 256-bit AES. Each offers you a range of options to protect data using either server-side or client-side encryption.

Google Cloud Key Management and AWS Key Management Service (KMS) are the competing encryption services on offer. Each offers you the capability to easily create and manage the keys used to encrypt and digitally sign your data.

Firewalls

Acting as the first line of defense for your IT infrastructure, a firewall is responsible for protecting your network from unwanted intrusion. Both Google Cloud and Amazon deliver state-of-the-art firewall protection of their cloud platforms.

In addition to this, both providers offer firewall-as-a-service products to enhance protection if you operate a Virtual Private Cloud (VPC), defend against DDoS attacks, and centralize the management of your firewall setup.

AWS Network Firewall and Google Cloud Firewalls are the competitive services that allow you to deploy network security access across your VPCs in just a few clicks. If you’re after protection against DDoS attacks, you can choose between similarly named services of AWS Shield or Google Cloud Armor.

When it comes to the central configuration and management of firewall rules across your cloud-hosted accounts and applications, Amazon offers this as a separate service named AWS Firewall Manager. Features and functionality that Google includes as part of its core Cloud Firewall service.

Identity Access Management (IAM)

Controlling who has access to what plays a critical role in system security, this is widely known as Identity Access Management. It’s the first step in preventing unwanted visitors from gaining access to sensitive information.

Both Google Cloud and AWS deliver Identity Access Management services within their cloud platform. Giving you granular control over who has access to your applications, what data they can access, and what they can do to your data.

Below are the core competing IAM services on offer:

Identity and Access Management Console

This is the centralized IAM service giving you complete visibility and control to manage your cloud resources. Giving administrators the power to control who can take action on specific resources.

Managed Services for Microsoft Active Directory

If you already implement IAM using Microsoft’s Active directory and plan to continue doing so in the cloud, both cloud providers offer you a hardened service for running Microsoft AD.

Single Sign-On

Centrally manage and control user access to multiple accounts and applications both on-premises and in the cloud through single sign-on access. Helping you deliver improved productivity and UX of employees through the ease of access.

Mobile and Web Application Control

Take advantage of a cloud-based IAM service letting you add user sign-up, sign-in, and access control to your mobile and web apps.

Shared Responsibility

Implementing security and compliance in the cloud is a shared responsibility.

It’s vital you understand the divide between who is responsible for what when it comes to implementing a robust cloud security stance. A misunderstanding here will create security vulnerabilities that are easily avoidable.

Amazon Web Services and Google Cloud Platform provide comprehensive guidance on their shared responsibility models for cloud security. Below is a high-level graphical depiction of each.

AWS Shared Responsibility Model

Amazon AWS Shared Responsibility Model
AWS Cloud Platform Shared Responsibility Model (Image Source: AWS)

Google Cloud Shared Responsibility Model

Google-Cloud-Platform-Shared-Responsibility-Matrix
Google Cloud Platform Shared Responsibility Model (Image Source: Google)

Support

Deploying a new cloud service, you are going to come across instances where you lack the prerequisite knowledge or expertise to achieve a task. In these situations, you want a cloud provider who has the additional guidance and support you need to overcome such obstacles.

Both AWS and Google Cloud are renowned for their extensive libraries of technical documentation. As well as their own thriving cloud communities, packed with 1000s of cloud experts who are always willing to share their knowledge.

Here you can browse a wealth of topics spanning tutorials, discussions, and even in-person meetups. Below you can find links to respective documentation and community support portals:

You can resolve most of the issues you run into with support from the sources above. Eventually, you will run into a situation where immediate, advanced expertise, and hands-on support are required. In this situation, it’s sensible to have an official support solution in place, direct from your cloud provider or an authorized third-party partner.

As part of their support model, both AWS and Google Cloud offer basic support. Alongside a range of additional paid premium plans. If you’re considering a premium plan, research and understand what’s included. This involves the associated fees, to ensure you pick a plan you need with a price you can afford.

Google Cloud Support Plans

Google Cloud has 4 available support plans, which split into two types – role-based support and premium support.

Role-based support is divided into three tiers – Basic, Development, Production:

  • Prices range from free to $250/month per user.
  • Each additional tier increase provides more support types, accelerated response times, more communication channels, greater availability, and escalation choices for more immediate issues.
  • Development and Production support plans can be combined for maximum coverage.

Premium support is the highest level plan available:

  • Prices can be upwards of $150k/year with an additional 4% of GCP and/or Google Workspace spend
  • You’ll receive guaranteed support response times within 15-minutes, 24/7 support for critical-impact issues, a Technical Account Manager, intelligent support systems, and even training.
  • The plan is fully customizable, allowing you to tailor support across the products and services most important to your organization.

AWS Cloud Support Plans

AWS also has 4 available support plans, which are split between free and premium.

Premium support is divided into 3 tiers – Developer, Business, and Enterprise:

  • Prices start from $29/month + 3% of AWS usage and scale upwards of $1200/month based on a percentage of monthly AWS usage which declines the more you spend
  • Each additional tier levels up your support with best practice checks, additional communication channels, 24/7 availability, issue response times within 15 minutes for business-critical system outages, a support API, a Technical Account Manager, and training resources.
  • Higher-level plans are also customizable, allowing you to pick and choose the products and services for which you desire premium support.

Here at Kinsta, we understand the importance and need for expert support in times of need. That’s why the entire Kinsta support team is made up of expert WordPress and Linux engineers. Whether you’re an SME or Fortune 500 company, you’ll get the same level of dedicated premium support.

Billing and Pricing

Without a doubt, accurate pricing comparisons between cloud providers is one of the most challenging aspects of the decision process. Each provider has a unique billing and pricing methodology with countless variables and moving parts.

To help you better understand the challenge of cloud provider pricing comparisons, below are just a few variables that will influence pricing up your desired cloud deployment:

  • Virtual Machines – number of instances, Ram requirements, number of CPUs, reserved or temporary instances
  • Storage Disks – storage amount required, data types, redundancy requirements, network-attached or locally-attached
  • Subscription model – whether you’re purchasing by the second, minute, hour, day, month, or year
  • Support – which tier you opt for, whether you customize your support, your average monthly cloud spend
  • Payment model – whether you’re selecting a pay-as-you-go service, reserved instance, or long-term committed use contract
  • Location – data center location also influences pricing

The larger your cloud deployment, the greater the complexity. Especially when factoring in the different technology types between cloud providers. Take VMs for example, the differing technology may make it impossible for a like-for-like comparison for RAM and CPU requirements.

Fear not, we’ve got some tools, information, and guidance to help you get started in creating your own personalized Google Cloud vs AWS pricing comparison.

AWS vs Google Cloud Pricing Comparison

There are literally 100s of different products available from Google Cloud and AWS. Each with its own sub-set of services, technologies, and pricing models. The options available mean deployment combinations can easily tick over into the 1000s. It’s unsurprising many are overwhelmed, even exploring storage and compute combinations for the most basic deployment.

Cloud Pricing Calculators

Thankfully, both providers each have their own comprehensive pricing calculator. It contains every product and service, specifications, and associated costs. This is your first step in creating a comparable pricing estimate.

For the purpose of this pricing comparison, we’ll explore VM compute costs from Amazon EC2 and Google Compute Engine. We’ve chosen this comparison option as, according to Gartner, two-thirds of total cloud spend is typically on compute resources. Also, in most instances, compute resources will form the foundation of your cloud deployment. So, without further delay, let’s take a deep-dive.

Assumptions for Cloud Pricing Comparison

In order to create an accurate comparison, we’ll select the same region, CPUs, and operating system for our compute setup:

  • Region: Northern Virginia – US East
  • Operating System: Linux
  • vCPUs/Cores: 4

We’ve then selected VM instances with comparable RAM specifications across the different machine types:

  • General Purpose
  • Compute Optimized
  • Memory-Optimized
  • GPU instances/VMs

Feel free to play with your options, as you’ll find switching between different variables of instance types, region, operating system, and CPUs can significantly alter your price per hour.

Below is a table outlining the chosen instances for comparison:

Instance Type Amazon EC2 EC2 RAM(GiB) Compute Engine Google RAM
(GiB)
General Purpose t4g.xlarge 16 n1-standard-4 15
Compute Optimized c6g.xlarge 8 c2-standard-4 16
Memory-Optimized r6g.xlarge 32 n2-highmem-4 32
GPU g4dn.xlarge 16 NVIDIA® Tesla® T4 64

Pay-As-You Go

AWS and Google Cloud offer an on-demand pay-as-you-go pricing model. This is best suited to individuals expecting intermittent cloud usage, as it allows you a flexible approach to add and remove services when you need them. Of course, this level of flexibility comes at a cost, making the pay-as-you-go model the most expensive per hour.

Instance Type Amazon EC2 EC2 Price
(per hour)
Compute Engine Google Price
(per hour)
General Purpose t4g.xlarge $0.134 n1-standard-4 $0.150
Compute Optimized c6g.xlarge $0.136 c2-standard-4 $0.188
Memory Optimized r6g.xlarge $0.201 n2-highmem-4 $0.295
GPU g4dn.xlarge $0.526 NVIDIA® Tesla® T4 $1.40

Table showing pay-as-you-go hourly rates of Amazon EC2 vs Compute Engine

As you can see from the table above, Amazon EC2 offers a significantly lower price per hour across the different instance types versus Google’s Compute Engine. This fact is increasingly impressive when you consider Compute Engine’s price per hour factors in a Sustained Usage Discount. This discount type is applied when usage in a month is above a certain threshold, offering savings starting at 15% and scaling up to 60%.

If you’re only seeking out compute resources for short intermittent periods it’s worth exploring temporary instances.

Referred to as Spot Instances by Amazon and Preemptible Virtual Machines by Google, you can unlock significant pay-as-you-go cost savings up to 90% of on-demand pricing above by tapping into the cloud providers’ spare compute resources.

As long as your happy for workloads to be interrupted if the resource is suddenly needed elsewhere.

Long Term Commitment Plans

If you’re planning for the long term—and can make a long-term upfront commitment to your cloud deployment – you will unlock significant savings versus a pay-as-you-go model.

Amazon and Google both offer a long-term pricing model with upfront commitment options of 1 or 3 years. Google named its plans Committed Use, while Amazon uses the term Reserved Instances. Both offer a significant discount over on-demand pricing. Up to 70% on Compute Engine and up to 72% on Amazon EC2.

Again, play with the variables to meet your needs – region, instance type, CPUs, operating system – as it will all influence your price per hour. With Amazon EC2, you’ll also find you can influence the discount amount based on when and how you pay.

There is also the option to choose convertible instance types, allowing you to switch to a newer VM if it becomes available.

Non-convertible instances, with the full amount paid upfront, offer the greatest level of discount. For the purpose of this comparison over a 1 and 3-year commitment, we’ve used these options.

1-Year Commitment

As you will see from the table below, again Amazon EC2 is cheaper across the board for a 1-year committed instance versus Compute Engine.

Instance Type Amazon EC2 EC2 Price
(per hour)
Compute Engine Google Price
(per hour)
General Purpose t4g.xlarge $0.079 n1-standard-4 $0.125
Compute Optimized c6g.xlarge $0.080 c2-standard-4 $0.141
Memory Optimized r6g.xlarge $0.118 n2-highmem-4 $0.177
GPU g4dn.xlarge $0.309 NVIDIA® Tesla® T4 $0.880

Table showing hourly rates for a 1-year commitment to Amazon EC2 vs Compute Engine

Amazon EC2 is up to 40% cheaper across the board for a 1-year committed instance versus Compute Engine.

The price gap widens with Amazon offering a greater discount of 40% against pay-as-you-go models for all instance types, as a reward for commitment. Whereas Google’s discount reward for your loyalty only amounts to 15-20%.

It’s important to note, by switching to no upfront payment and the convertible instance option on Amazon EC2 your discount amount drops below 30% and narrows the price difference somewhat.

3-Year Commitment

Looking at the table below, we’re experiencing déjà vu, with Amazon EC2 continuing to be cheaper across the board when making a 3-year commitment versus compute engine.

Instance Type Amazon EC2 EC2 Price
(per hour)
Compute Engine Google Price
(per hour)
General Purpose t4g.xlarge $0.050 n1-standard-4 $0.046
Compute Optimized c6g.xlarge $0.051 c2-standard-4 $0.094
Memory Optimized r6g.xlarge $0.075 n2-highmem-4 $0.126
GPU g4dn.xlarge $0.198 NVIDIA® Tesla® T4 $0.640

Table showing hourly rates for a 3-year commitment to Amazon EC2 vs Compute Engine

However, we have an exception in the General Purpose category, Compute Engine bucks the trend and is cheaper under a 3-year commitment.

Otherwise, the price gap remains around 40% for Compute Optimized and Memory Optimized options over a 3-year commitment. For GPU comparison, it actually widens to around 60%, a massive saving.

Again, I should note, switching to no upfront payment and the convertible instance option on Amazon EC2 will have a more pronounced reduction on your discount amount. This reflects the increased risk of keeping VM instances over 3 years.

Cloud computing evolves fast, being tied in will prevent you from tapping into newer, faster, and more efficient VM instances.

Free Trials

If you’re not ready to make the switch to a cloud service, both AWS and Google Cloud offer the option of a free tier across a wide range of their products. Giving you a predefined resource amount over a set amount of time, perfect if you’re looking to trial a service.

Both providers also offer ‘always free’ cloud services, which are ideal if you have very low usage requirements and don’t mind operations being interrupted. Let’s take a closer look.

AWS Free Tier

Exploring the AWS Free Tier, you’ll unlock free access to a range of 85 cloud products and services.

The AWS Free Tier has three different types:

  • Always free – a free offer that never expires and is available to all AWS customers
  • 12 months free – available free over the first 12-months from your initial sign-up to AWS
  • Trials – free for a shirt-term following the activation of a particular service

You’ll be able to explore a wide range of products across compute, storage, database, IoT, AI, and many more.

If you’re starting out then it’s worth considering the compute and storage options which are free for 12-months after signing up:

  • Compute – Amazon EC2 access for 750 hours a month with a t2. or t3. micro instance
  • Storage – Amazon S3 standard storage of 5GB per month with 2,000 put 20,000 get requests

Google Cloud Free Tier

Exploring the GCP Free Tier appears a little more restrictive than the AWS alternative. Though you’ll still get access to 24 cloud products and services. Which unlike AWS, remain under and always free offer, within monthly usage limits of course.

In addition to AWS’s free offerings, new Google Cloud customers will unlock $300 of free credit which can be spent on ANY of the Google Cloud products and services.

Although your options are more limited, you can still explore an exciting array of products across IoT, AI, storage, database, and compute that will largely cover the most commonly desired cloud services.

Much like with AWS, if you’re starting out, trial the compute and storage options which remain always free on the GCP:

  • Compute – Compute Engine access to an F1-micro instance with 30GB HDD per month and a 5GB snapshot
  • Search – Cloud Storage availability of 5GB for standard storage with 5,000 put and 50,000 get requests

When it comes to comparing the free tiers, it’s clear GCP has an edge over the AWS platform. Like for like, they will provide you with much wider access to trialing their different products and services. Ideal if you’re not yet ready to make a commitment to a cloud deployment.

Is Google Cloud Cheaper than AWS?

When it comes to cloud compute resources, which form the backbone of most cloud deployments and spend, the answer is no. AWS is definitely cheaper than the Google Cloud Platform for VM instances.

However, the answer becomes a lot more “cloudy” when you move away from simple compute resources. After researching multiple products, services, and pricing models there is no clear winner in the price war.

You’ll need to find your own answer to this question, and it will hinge entirely on the unique requirements of your business. Which data center location you chose, your networking requirements, the type of workloads you’re running, seasonality. The list is never-ending.

One thing is for sure, there are deals to be had, and the opportunity for Google Cloud to be cheaper than AWS certainly exists. It just depends on your setup and the services required.

To add Azure to the mix, check out our cloud computing comparison of AWS vs Azure.

Summary

Our goal of researching these cloud providers is in seeking the definitive answer to which is the better cloud platform, Google Cloud vs Amazon Web Services?

On our journey, it’s clear Google Cloud and AWS are market leaders. Both platforms offer an extensive range of cloud products and services at the cutting edge of technological advancement. Giving you significant benefits over an on-premises deployment when it comes to scalability, performance, security, and cost. Whoever you chose, you’ll unlock a premium service at a competitive price.

At Kinsta, we use the GCP platform to deliver top-tier performance within our web hosting solutions. The premium tier network service offers significant performance enhancements in reducing latency and minimizing downtime. In terms of progression, it’s clear that Google Cloud is making significant improvements to the platform.

In the end, the answer to which is better depends on the individual needs of your business. Whatever that answer looks like, do your research, follow our Google Cloud vs AWS guide and make the best decision for your business.

Edward Jones

Edward Jones is a technology writer with 8 years of industry experience. He has published over 300 articles with major publications that include Microsoft, IBM, and Entrepreneur.