Google Cloud Review

Compute Engine is the primary compute service Google Cloud offers. It is an Infrastructure-as-a-Service (IaaS) offering, so Platform-as-a-Service (PaaS) products like Looker and GKE are built on it.

Compute Engine can be used for anything you’d use any other server for. It offers various instance types, from general-purpose machines to storage-optimized, compute-optimized, memory-optimized and accelerator-optimized machines, each one suited for varying computing needs.

Google Kubernetes Engine is Google Cloud’s managed Kubernetes platform. It is a Platform-as-a-Service (PaaS) tool that functions essentially the way the regular Kubernetes does, but with more abstraction. For one, you won’t have to set up a cluster from scratch with a tool like kubeadm — GKE will do most of it for you.

Like Kubernetes, GKE is used for container orchestration. It manages and monitors containers and containerized applications throughout their life cycle.

Speech-to-Text is an AI-powered voice transcription service and is a marker of Google Cloud’s excellence in the AI sector. The primary function of Speech-to-Text is to transcribe vocalized speech into written text.

Beyond transcription, Speech-to-Text offers accent and language recognition through its Chirp model — one of its pre-existing models. That said, you can also create custom models. Speech-to-Text currently supports 125 languages. 

Looker is a PaaS tool used for enterprise-level data analysis and business intelligence. It also offers comprehensive dashboards and API integrations for developers.

Looker comes with a 30-day free trial, but if you prefer a completely free tool, you can use Looker Studio. Looker Studio offers a platform where you can analyze and visualize data yourself. It has no paid features like custom applications or AI-driven data analysis.

BigQuery is Google Cloud’s data analytics service; it is a data warehouse that primarily stores big data. Beyond data storage, BigQuery processes and analyzes data in various ways, including data cleaning, predictive analytics, real-time analytics and marketing analytics.

Cloud Functions is analogous to AWS Lambda. It is a serverless service that runs codes/functions in response to an event or invocation. Though referred to as a serverless service, Cloud Run does use servers, but those servers are abstracted from you.

You can use Cloud Run functions for file transformations, file analysis, mobile backends or IoT backends. Cloud Functions executes transient codes, so they only run for a short period — never longer than 9 minutes for event-driven functions or 60 minutes for HTTP functions. 

Google Cloud SQL is a managed relational database service that offers databases in three database management systems: MySQL, PostgresQL and SQL Server. As a managed service, processes like data replication, patching, encryption and backups are automated, so you don’t have to worry about them.

Google Cloud promises 99.95% SLA for general workloads and 99.99% SLA for intensive workloads on Cloud SQL Enterprise, which is pretty impressive.

Cloud Storage is GCP’s object storage service — Google Cloud’s version of AWS’ Amazon S3 and Azure’s Blob Storage. It offers four storage classes: Standard, Nearline, Coldline and Archive.

Standard is suited for frequently accessed data, while Archive is great for infrequently accessed data (data accessed less than once every 365 days). Like Archive, Nearline and Coldline are for infrequently accessed data, with minimum storage periods of 30 and 90 days, respectively.

When it comes to storage pricing, Standard is the most expensive, while Archive is the cheapest. That’s because data in Standard is accessed more frequently, so it has higher availability, whereas the opposite is true for Archive. On the other hand, Archive comes with higher costs for data access and operation.

Virtual Private Cloud, or VPC, is a private network service within Google Cloud. It offers a way to isolate resources for a project in your Google Cloud account, ensuring security and privacy.

Cloud VPC offers features like routing, a firewall, VPC peering, a VPN and packet mirroring. The routing feature allows you to control the flow of traffic between instances within the VPC, and the firewall restricts access to your instances. VPC peering allows communication between different Cloud VPCs, and the VPN facilitates connecting non-Google products with Cloud VPC.

Cloud Monitoring is a platform that offers insight into the activity, performance, health and availability of your infrastructure and applications. It allows you to monitor your resources based on preexisting and custom metrics. It also generates dashboards, offering a consolidated view of the data you’re monitoring.

If you’re familiar with Prometheus, you can get a managed Prometheus service using Google Cloud. Additional integrations include alerting, logging and Google Cloud console integration.

Google Cloud offers three main types of storage services: object, file and block storage via the Cloud Storage, Filestore and Block Storage services, respectively. Cloud Storage offers 5GB free per month to users in U.S. regions. However, beyond that limit you will be charged based on your storage class.

Cloud Storage Standard costs $0.02 per GiB per month, while Nearline, Coldline and Archive cost $0.01, $0.004 and $0.0012 per GiB per month, respectively. Besides paying for storage, you’ll pay between $0.12 and $0.20 per month for general network usage, depending on your monthly usage. Data transfer charges within GCP depend on the networking service used.

Filestore pricing depends on the instance capacity, instance service tier and region. For example, in the us-east1 region, you’ll pay $0.30 per GiB per month for a Basic SSD Premium service tier instance. Therefore, if your instance capacity is 20 GiB, you’ll pay $6 per month, even for partial usage. The same instance in europe-west1 will cost you $6.60 per month.

Block Storage pricing varies across types, but Hyperdisk storage is generally charged based on IOPS (input/output operations per second), throughput and capacity. Hyperdisk capacity charges vary from $0.005 to $0.125 per GB of provisioned capacity. Throughput Standalone is the cheapest and Hyperdisk Extreme is the most expensive.

The throughput charge with Hyperdisk Throughput is $0.25, and with Hyperdisk Balanced it is $0.04. The Hyperdisk Balanced IOPS charge is $0.005 per provisioned IOPS and $0.032 for Hyperdisk Extreme.

A local SSD costs $0.036 for provisioned capacity, and Persistent Disk costs between $0.04 and $0.125 for provisioned capacity. You’ll also pay $0.065 for standard Snapshot storage or $0.019 for Snapshot archive storage and retrieval.

Cloud Storage usage costs primarily consist of egress costs (cost of data transfer out of Cloud Storage) and operations costs. 

Egress Rates

Data transfer into Cloud Storage — ingress — comes at no cost; however, outward data transfer — egress — may involve a cost. Generally, data egress within a Cloud Storage bucket region is free. Egress outside a bucket’s region comes at a cost, depending on the destination region and the bucket region. Egress rates fall between $0.02 per GB and $0.14 per GB.

Egress outside Google Cloud costs around $0.08 to $0.23, depending on the origin and the destination. The rate is cheaper with larger data transfers.

Operation Rates

Operations in Cloud Storage are divided into class A, class B and free. Class A operations cost $0.005 per thousand operations for Cloud Storage’s Standard storage class in a single region and $0.01 for the same class in a multi-region. Nearline storage and Durable Reduced Availability (DRA) storage cost $0.01 in a single region and $0.02 in a multi-region.

Coldline storage goes for $0.02 per 1000 class A operations in a single region and $0.04 in a multi-region, while Archive costs $0.05 and $0.10 in a single region and a multi-region, respectively.

For class B operations, Standard storage, Nearline storage and Durable Reduced Availability (DRA), Coldline, and Archive cost $0.0004, $0.001, $0.01 and $0.05, respectively, in both a single region and a multi-region.

Google Cloud Platform offers various always-free services; however, these services are free only within certain usage limits. For example, you get only one e2-micro instance for free per month with the Compute Engine service. For clarity, your total usage of e2-micro instances in eligible regions should not exceed the total monthly hours of one e2-micro instance.

Other free services from GCP include BigQuery (1TB of queries per month), Cloud Functions (2 million invocations per month), Speech-to-Text (60 minutes free per month), Cloud Storage (5GB free per month in U.S. regions) and Google Kubernetes Engine (one zonal or autopilot cluster per month).

Besides these always-free services, Google Cloud issues free trials for services like Google Kubernetes Engine Enterprise, Looker, Spanner and AlloyDB.