AWS vs GCP Which is Better in 2021

AWS vs GCP Which is Better in 2021 | Networking Funda

  • Post comments:6 Comments
  • Reading time:13 mins read

Only a few years in the past, enterprises have been reluctant to migrate to the cloud. Now, the cloud is a mainstream era, multi-cloud is the wave of the future, and hybrid environments are gaining in popularity. Flexera’s state of the Cloud 2019 survey4 observed that.

  • 94% of enterprises use the cloud, and 31% say that public cloud is their #1 priority.
  • Multi-cloud is by far the preferred strategy; 84% of enterprises run multiple clouds.
  • 58% of enterprises run hybrid environments, up from 51% the year prior.

Organizations now find themselves tasked with evaluating and selecting not just one but multiple public cloud providers, then integrating and maintaining them, along with on-prem infrastructure.

Before we get deep dive into AWS vs GCP which is better in 2021, let’s go through the Overview.

AWS vs GCP Overview:

Launched in 2006, AWS was one of the first pay-as-you-go cloud computing models to be offered to the general public. Google launched GCP in 2008. Comparisons of AWS and GCP frequently claim that the public cloud is a “new” venture for Google.


While it’s true that AWS has been selling cloud services to the general public for a long, Google is not “new” to the cloud. In fact, Google’s cloud infrastructure predates Amazon’s.

Google developed Borg, the predecessor to Kubernetes, in about 2003 or 2004, and used it to manage production containers internally before introducing the open-source Kubernetes in 2014 and officially releasing it in 2015.

The company needed a highly secure and massively scalable platform for its ambitious
internal projects, including Google Search, Maps, AdSense, and Gmail.


At the time, no other company was doing what Google was doing — and certainly not on the same scale — so Google had to build its own solution! GCP simply allows other enterprises to take advantage of the same secure, time-tested, and highly optimized cloud infrastructure that Google has relied upon for years.

AWS vs GCP Service Comparison:

Both GCP and AWS offer a core set of services for compute, storage, networking, and databases. Higher-level services, such as machine learning and application services are built atop these core features:


Compute: Google Compute Engine and Google App Engine | Amazon Elastic Compute Cloud (EC2).


Storage: Google Cloud Storage | Amazon Simple Storage Service (S3) and Amazon Elastic Block Store (EBS).


Networking: Google Virtual Private Cloud | Amazon Virtual Private Cloud (VPC)


Databases: Google Cloud SQL, Google Cloud Firestore, and Google Cloud Bigtable | Amazon Relational Database Service (RDS) and Amazon DynamoDB.


This table provides a side-by-side comparison with more detail:

Compute:

Service CategoryServiceAWSGoogle Cloud
ComputeIaasAmazon Elastic Compute CloudCompute Engine
PaasAWS Elastic BeanstalkAPP Engine
ContainersAmazon Elastic ContainerGoogle Kubernetes Engine
Containers without InfrastructureAWS FargateCloud Run
FaasAWS LamdaCloud Functions
Managed Batch ComputingAWS BatchN/A

Network and Storage:

Service CategoryServiceAWSGoogle Cloud
NetworkVirtual NetworkAmazon Virtual Private CloudVirtual Private Cloud
Load BalancerElastic Load BalancerCloud Load Balancing
Dedicated InterconnectDirect ConnectCloud Interconnect
Domains and DNSAmazon Route 53Google Domains, Cloud DNS
CDNAmazon CloudfrontCloud CDN
StorageObject StorageAmazon Simple Storage ServiceCloud Storage
Block StorageAmazon Elastic Block StorePersistent Disk
Reduced-availability storageAmazon S3 Standard
Infrequent Access, Amazon S3
One Zone-Infrequent Access
Cloud Storage Nearline and Cloud Storage Coldline
Archival StorageAmazon GlacierCloud Storage Archive
File StorageAmazon Elastic File SystemFilestore

Database and Big data Analytics:

Service CategoryServiceAWSGoogle Cloud
DatabaseRDBMSAmazon Relational Database Service, Amazon AuroraCloud SQL, Cloud Spanner
NoSQL: Key-valueAmazon DynamoDBFirestore, Cloud Bigtable
NoSQL: IndexedAmazon SimpleDBFirestore
Big Data & AnalyticsBatch Data ProcessingAmazon Elastic MapReduce, AWS BatchDataproc, Dataflow
Stream Data ProcessingAmazon KinesisDataflow
Stream Data ingestAmazon KinesisPub/Sub
AnalyticsAmazon Redshift, Amazon AthenaBigQuery
Workflow orchestrationAmazon Data Pipeline, AWS GlueCloud Composer

Management Services:

Service CategoryServiceAWSGoogle Cloud
Management ServicesMonitoringAmazon Cloud WatchStackdriver Monitoring
LoggingAmazon Cloud Watch LogsStackdriver Logging
DeploymentAWS Cloud FormationCloud Deployment Manager

Machine Learning:

Service CategoryServiceAWSGoogle Cloud
Machine LearningSpeechAmazon TranscribeSpeech-to-text
VisionAmazon RekognitionCloud Vision
Natural language processingAmazon ComprehendCloud Natural Language API
TranslationAmazon TranslateCloud Translation
Conversational InterfaceAmazon LexDialogflow Enterprise Edition
Video intelligenceAmazon Rekognition VideoVideo Intelligence API
Auto-generated modelsN/AAutoML (Beta)
Fully Managed MLAmazon SagemakerAI Platform

What makes GCP the superior option?

Cost savings through pricing innovations!


AWS bills are notoriously complicated and filled with hidden costs, such as unused or underutilized EC2 instances. Ever-increasing AWS bills were a major reason why ME.ME, a search engine for memes switched from AWS to GCP. 


Reaching out to AWS didn’t help the company bring down its tab. “I never spoke to anyone on the phone, just through online chat, and no one on their side ever told us how to cut costs,” says Jim Hefner, ME.ME’s CTO.


In contrast, GCP strives to give users as much visibility into their cloud costs as possible, along with easy-to-use cost optimization tools that help users keep spending under control.


On average, GCP customers can save 21% over AWS on online storage workloads; additionally, GCP offers automated right-sizing recommendations, sustained-use discounts, and other cost-saving tools to save users an average of 35% on computing workloads.

Sustained use discounts:

Sustained use discounts are a feature unique to GCP. These discounts, which are based on a sliding scale according to percentage usage, are automatically applied each month.

They do not require prepayments or commitments, and users may combine non-overlapping instances (“inferred instances”) to maximize their discount.

GCP users can save up to 30% on workloads that run for a significant portion of the billing month on Compute Engine and Cloud SQL.

Big savings with microservices:

GCP allows for the abstraction of cloud technologies from memory-sucking virtual machines to modern platforms that facilitate “just right” microservices that significantly reduce wasted cloud spend.

For example, instead of running 400 virtual machines, each with 75% utilization (the equivalent of 100 of those VMs going unused), GCP users can deploy 4000 Docker containers running in perfect orchestration via Google Kubernetes Engine, each with 95% utilization.

Fixed-price preemptible virtual machines:

GCP’s preemptible VMs (PVMs) lets users save up to 79% on workloads that can be interrupted, such as data mining and data processing.

Unlike AWS Spot Instances, which work on a dynamic pricing model, GCP PVMs are fixed-price, so organizations can better predict their costs.

Significantly lower TCO for EDWs:

Enterprise Strategy Group (ESG) conducted a three-year total-cost-of-ownership (TCO) study that compared upgrading an on-premises enterprise data warehouse (EDW) solution from a leading vendor, migrating to a cloud-based solution provided by the vendor on AWS, or redesigning and migrating to Google BigQuery. 


In the end, Google was the clear winner; ESG found an overall three-year cost reduction of 52% compared to remaining on-premises, along with a 41% reduction compared to using the vendor’s solution on AWS.

Custom machine types:

GCP users can choose any configuration of CPU and memory to save up to 48% compared to fixed machine types from other cloud providers.

Get an accurate comparison of your cloud spends:

A line-by-line comparison of cloud providers’ costs across all services is a highly complex undertaking that is outside the scope of this paper. All vendors use different terminology to describe their products, instance types, and pricing plans. 


Additionally, variables such as load balancing, networking, on-demand pricing, and committed-use discounts can significantly impact costs. Organizations can generate high-level estimates using Google’s online pricing calculator.


However, aS far more efficient and accurate method is to use the free CloudPhysics tool, which assesses organizations’ current infrastructure then provides a side-by-side comparison of their current spend with what they would be spending on GCP.

In addition to AWS, this tool also works with Azure and on-prem hardware.

Simpler, more flexible service usage tracking & billing:

GCP and AWS organize service usage tracking and billing quite differently. AWS tracks all usage according to a user account; any services utilized are billed to the account used to sign up for AWS.

Users can also create billing accounts, then create sub-accounts whose usage rolls up to them. Tracking who is responsible for what portion of the bill can get quite complicated when dealing with multiple divisions, project teams, and other groups.


Conversely, GCP organizes service usage and billing according to the project, not the user account. Users can create multiple and completely separate projects under the same account.

Billing is accomplished through billing accounts, and individual billing accounts can be linked to one or more projects. 

This model allows enterprises to easily create dedicated project spaces for separate divisions, locations, or other groups within a company, with all charges for that group, billed on one invoice.


The ability to create siloed project spaces is also helpful when conducting testing. A user can create a test project, then delete it when they’re finished, ensuring that all the resources created for the test are also deleted without impacting other projects.


Game developer FlowPlay appreciates the flexibility of paying for computing power instead of hardware: “AWS requires customers to choose a specific server configuration when buying a long-term contract,” explains Douglas Pearson, Flowplay’s Co-Founder and CTO. “With GCP, instead of renting a specific server, we buy computing power.

This allows us to experiment with servers, drives, configurations, and RAM to optimize performance. We couldn’t have done that as easily with AWS.”

Robust artificial intelligence/ machine-learning tools:

AI/ML capabilities are a major factor when choosing a public cloud provider. IDC predicts that by 2022, 75% of enterprises will be using AI solutions to analyze data and glean actionable, innovation-driving business insights.


GCP’s AI/ML and data analytics capabilities were a big selling point for Northgate Market, an ethnic foods grocery chain. “Those are core strengths for Google that AWS doesn’t have,” says Harrison Lewis, Northgate’s Chief Information and Privacy Officer. “We knew we wanted to build a data lake on BigQuery to manage our customer data; that was an important feature for us.


Startup food delivery platform FoodJets is making heavy use of GCP’s ML features to enhance their end users’ experience. “We’re using AutoML because we don’t have data scientists on our team. We can have our developers train models, then use trial and error to refine them. It’s very powerful and easy to use,” explains CTO Veer Singh.


Google has long been committed to research in AI/ML. Google’s research division, Google AI, employs a team of engineers devoted to using AI/ML to solve both internal business problems and big-picture societal issues.

Its engineers frequently author academic research papers to publicly share their findings, and the AI/ML tools available in GCP are the same as that Google uses in-house.

Comprehensive, easy-to-configure cybersecurity:

Cybersecurity is another core competency for Google, born of necessity. The world’s most popular search engine is also the world’s biggest cyberattack surface. Every minute of every day, Google’s cybersecurity tools: 


Prevent 10 million spam messages from reaching Gmail customers.


‐ Scan 694,000 indexed Web pages for harmful software. 
‐ Intercept and stop 7,000 deceitful URLs, executables, and browser extensions that may carry viruses, unwanted content, or phishing attempts. 
‐ Report 6,000 instances of unwanted software and nearly 1,000 instances of suspected malware to Chrome users. 
‐ Identify and label two phishing sites and one malware site.


To accomplish these tasks, Google’s cybersecurity engineers must have a deep understanding of the real-time threat environment. Hundreds of the world’s leading experts in information, application, and network security work to protect the GCP infrastructure.


Cyber-attacks due to misconfigured cloud settings are at epidemic levels. According to McAfee, while organizations estimate they average 37 IaaS misconfiguration issues per month, the actual number can approach 3,500.38 To stem this tide, Google has gone out of its way to make GCP’s security controls as easy to use as possible.

Kubernetes expertise from the developers of Kubernetes:

Kubernetes is one of the world’s most popular container orchestration tools, and it’s only getting more commonplace. Flexera found that Kubernetes adoption nearly doubled between 2018 and 2019, skyrocketing from 27% to 48%.


AWS offers Kubernetes services, but Google is the creator of the Kubernetes project and remains the dominant contributor to its codebase. This “home-field advantage” makes GCP a particularly attractive choice for DevOps organizations. 


GCP users get to access new Kubernetes features and deployments immediately, while rollouts on AWS are delayed. They also get the networking stack that Kubernetes was designed to operate on, which simplifies configuration and operations.

Google Kubernetes Engine (GKE), widely considered the industry standard for running Kubernetes, automates baseline functionalities, while AWS requires a lot of manual work to set up Kubernetes clusters. 

This makes GKE more user-friendly than Amazon EKS, especially for developers who are new to Kubernetes or containers.


Container security and compliance can be tricky. Anthos Config Management, a key component of the Google Anthos hybrid and multi-cloud management solution automates policy and security for Kubernetes clusters at scale. 


Out of the box, users can create multi-cluster policies that set and enforce role-based access controls, resource quotas and create namespaces on all Kubernetes clusters, both on-prem and in the cloud. AWS has no equivalent feature. 

Reliability & performance:

Google’s global network consists of thousands of miles of fiber optic cable and it utilizes advanced networking and edge caching services to deliver fast, consistent, and scalable performance.

GCP has the largest private network of any public cloud vendor, with over twice the number of submarine cables as AWS


Because GCP’s points of presence (POPs) connect to Google’s data centers via Google-owned fiber, GCP-based applications have fast, reliable, and unimpeded access to all GCP services. GCP’s 99.95% SLA is in line with AWS.

However, unlike GCP, AWS’s SLA applies only to the availability of the control API, not individual VMs. 


Google Compute Engine offers live migration to keep virtual machine instances running even when a host system event occurs, such as a software or hardware update.

Conversely, AWS occasionally performs maintenance on the hardware that underlies EC2 instances, which might require a few minutes of downtime. 


This is why AWS will not commit to a 99.95% uptime SLA on virtual machines. Google’s 24/7/365 reliability was a major reason why game developer FlowPlay chose GCP.

In the online gaming industry, there is no such thing as ‘down for maintenance,’” notes Douglas Pearson, Flowplay’s Co-Founder and CTO. “There is no ‘offseason.’ There is no window where we can be offline, not ever.

Ease Of Use:

GCP is designed to be easy to use for both IT administrators and non-technical employees. Google has heavily invested in practical training that gets new users up to speed very quickly.

Coursera and Pluralsight offer on-demand training, and in-person classroom training are available around the globe.


The on-demand, entry-level GCP Fundamentals course can be completed in about one day.52 Cloud professionals who are experienced in AWS will find that much of their existing knowledge will transfer easily to GCP. the GCP training for AWS professionals is only six hours long.

Executive summary:

Google Cloud Platform (GCP) isn’t always just an alternative to AWS, but for lots of use cases, it’s far a much-advanced desire. 

GCP is greater value-effective1 and open2 than AWS and possesses deep knowledge in cybersecurity, artificial intelligence (AI)/device studying (ML), and container management.

“We chose Google Cloud Platform as it becomes the maximum dependable, fee-effective, and automatic cloud the solution to be had,” says Tim Morrow, CTO at TVG community.

“We get better safety, strong compliance, and the peace of mind that after the most important race day rolls around, we received’t have any downtime.

Also Read:

Google Cloud Platform Core Services | Networking Funda

Google Cloud Associates Exam Questions | Networking Funda

This Post Has 6 Comments

Leave a Reply