While Google is obviously the undisputed leader in the search engine market, in the cloud computing sector it’s coming from behind. Google Cloud Platform has changed leaders several times, and has had to adjust approach to better align with the cloud market.
So even though Google launched its cloud platform in 2008, just two years after Amazon Web Services, Google is lagging behind. In contrast to GCP, AWS has executed near flawlessly from its launch.
The market share numbers reflect this difference: Synergy Research Group puts GCP in fourth place with just 6% market share, which is dwarfed by AWS’s 34% market share. To put this in context: Microsoft Azure holds about 14% share, with IBM Cloud at 7%. Some research firms put China’s Alibaba Cloud ahead of GCP, at least in total sales, but that’s largely confined to Asia. Overall, these cloud vendors are the leaders among all cloud computing companies.
There’s little debate that AWS wins clearly in many cloud areas. It has an enormous lead over GCP in both quantity and quality of cloud services available. Additionally the AWS services are well integrated and have very good synergy.
However, it’s not wise to count out Google Cloud. In particular, GCP has deep and very well regarded expertise in artificial intelligence, machine learning and Big Data analytics. Arguably the company is the industry leader in these three data-intensive areas. Given the critical importance of these three data sectors, they certainly shore up the strength of GCP.
Additionally, Google’s resources are of course vast. The company has shown that it is willing to make course corrections and invest heavily in the cloud. GCP is “in it to win it,” so it’s reasonable to expect big steps forward in the months and years ahead.
The Core Platforms
Both companies come from a pure cloud play. Neither had an on-premises legacy like Microsoft, IBM, and Oracle. The two cloud companies have a lot in common and offer similar services, such as containers, DevOps, and databases. Both are often used as part of a hybrid cloud or a multicloud strategy.
But they differ, too. So let’s break it down.
AWS: The Basics
The basic Amazon AWS Infrastructure-as-a-Service (IaaS) offerings involves four segments:
- Content storage and delivery
Amazon Elastic Compute Cloud (EC2) and Simple Storage Service (S3) are the core offerings on which everything else is built. From there, Amazon goes into Elastic Beanstalk for its Platform-as-a-Service (PaaS) offering for building scalable web apps.
GCP: The Basics
GCP breaks down into five specific services:
- Computing and hosting
- Big data
- Machine learning
For basic IaaS, there is Google Compute Engine, which is a typical IaaS offering for rapid deployment of virtual machines. However, Google has a few advantages. It charges far less per virtual machine than any other provider and has much more fine-grained billing, down to the second, rather than the minute for AWS.
For PaaS, Google has Google App Engine, which allows for much greater scalability of its apps than AWS, but is limited in languages and operating environments. AWS supports Windows, while GCP is focused on NIX and open source languages.
AWS Pro and Con
Amazon has more than 100 cloud services, second only to IBM with 175, and IBM has that many because it brought much of its legacy software like DB2 and Watson to the cloud. AWS is mostly new offerings, except for databases and some tools.
AWS got its start with developers, and supporting DevOps remains integral to the company even as it has expanded its offerings. The basic tools is AWS Developer Tools, a set of four services for building AWS-hosted or on-premises apps. They are:
- AWS CodeCommit, to store code in a private Git repository.
- AWS CodePipeline, for continuous integration (CI) and continuous delivery (CD).
- AWS CodeBuild, to build and test the code.
- AWS CodeDeploy, to automate code deployments.
IDE toolkits include Eclipse and Visual Studio and a number of command line toolkits for automation of service management with scripts.
- AWS Command Line Interface
- AWS Tools for PowerShell
- AWS SAM Local
- Amazon EC2
- AWS Elastic Beanstalk
- Amazon EC2 Container Service
- Serverless Developer Tools
For lighter code development, like serverless computing, Amazon has AWS Lambda so you don’t have to provision virtual machines or servers. And if you are working on Infrastructure as Code (IAC) provisioning, which is even lighter weight than serverless, AWS has you covered as well through AWS CloudFormation for managing AWS resources.
Amazon has also done a lot in the area of analytics. AWS has a comprehensive set of analytics tools, such as Athena for analysis of data stored in S3 instances, EMR for Hadoop, QuickSight for business analytics, Redshift for a petabyte-scale data warehouse, Glue to perform ETL tasks on data stores, and Data Pipeline to securely move data around.
For BI, Amazon offers Redshift, a database deployed as a cluster for massive parallel processing for more advanced data warehouse users. For visualization of customer data, QuickSight is a business intelligence service. It uses data stored in any Hadoop repository, an Amazon RedShift data warehouse or third-party sources such as Salesforce.com and Oracle.
If there is a negative to AWS it is that for the longest time it operated on an assumption of a pure cloud play, that customers would bring everything to the cloud. That hasn’t happened. Multiple research reports indicate that most enterprises prefer a mixed hybrid environment. So now AWS is moving to extend its offerings on premises.
Also, with so many services, AWS’s cost structure is very confusing to navigate on your own, so to avoid billing surprises, it offers AWS CloudWatch, which offers metrics on data transfer, disk usage, and CPU utilization.
GCP Strengths and Weaknesses
GCP offers a number of IaaS/PaaS services, starting with compute:
- Compute Engine
- App Engine
- Kubernetes Engine
- GKE On-Prem – to make apps cloud-ready and move them to the cloud.
- Cloud Functions – serverless computing
- Knative – components to create Kubernetes-native cloud-based software.
- Shielded VMs – Hardened virtual machines on GCP.
- Container Security
It also has quite a collection of developer tools:
- Cloud SDK
- Container Registry
- Cloud Build – DevOps continuously build, test, and deploy.
- Cloud Source Repositories – A single place for your team to store, manage, and track code.
- Cloud Scheduler
- Cloud Tasks
- Cloud Tools for IntelliJ
- Cloud Tools for PowerShell
- Cloud Tools for Visual Studio
- Cloud Tools for Eclipse
- Gradle App Engine Plugin
- App Engine Plugin
- Firebase Test Lab
Many of the strengths of GCP originate with the Google search engine and its basic services. Google has a massive worldwide presence and is constantly building new, massive data centers all over the globe. GCP has what is called multi-regional deployment mode, so you can deploy your site in multiple regions for the fastest response times.
Google’s billing is considered a lot more customer friendly, with lower costs than the competition, discounts for long-running workloads and no up-front commitment required. AWS requires prepayment for reserved instances to be eligible for the discounts. It also offers a significant discount over AWS, as much as 50%, for SSD storage, which is much faster than hard disk storage, even in the cloud.
In addition to the infrastructure, Google has an unmatched network of fiber connections making its backbone faster than the Internet in some ways. It has cables between the U.S. and Europe and Asia with multi terabits of bandwidth. It is the only cloud provider with a tiered cloud network of standard and premium, so you can use Google’s network and not have to compete with Netflix and YouTube traffic.
One area where Google holds its own with AWS is security. All data is encrypted in transit between Google, the customers, and both of their respective data centers. All data in the Cloud Platform services and stored on persistent disks is encrypted, the latter under 256-bit AES encryption keys. Google has multiple layers of authentication.
But the big advantage for GCP is Google’s strong commitment to artificial intelligence and machine learning. Google has invested heavily in its own AI chip, called the Tensor Processing Unit (TPU). The TPU is not for sale and only available as a service on GCP. It is specifically designed for machine learning and is used by several Google services, including Translate, Gmail, and Search.
Google is currently on version 2 of the chip, which offers 180 teraflops of performance. Version 3, still in development, will offer 420 teraflops of performance.
AWS does have AI services called SageMaker and, as Google does with its own consumer services, they build on Amazon consumer services, such as Alexa. But to date AWS has nothing to match the performance of the Google TPU.
However, AWS Machine Learning is limited to a single type of ML model, logistic regression. While useful, it’s still limiting. GCP’s ML is more general purpose. Also, there is no way to export the models out of the service after they have been trained, which GCP does allow.
Finally, one strong point for GCP: Google developed Kubernetes, which is dominant in the container space. Even Microsoft supports Kubernetes on Azure. So GCP has the stronger container story.