Google Cloud Platform is still a relative newcomer to the enterprise cloud marketplace; the company only launched its IaaS offering, Google Compute Engine, as recently as 2013. But it has already emerged as a serious challenger to the dominant AWS and Microsoft Azure, boasting high-profile clients such as Apple, Spotify, Snapchat and Best Buy.
But if you’re thinking about using Google to host some of your enterprise IT workloads, you’ll need to think about the various options for storing your data.
This article covers the cloud vendor’s portfolio of data-storage services. In addition, to help users more familiar with the offerings of market leader AWS, where applicable, I’ve also listed the Amazon equivalent of each service. Finally, the article updates users about Google’s recently launched cold-storage service.
Persistent Disk is Google’s scalable block-storage solution, intended for use as attached volumes or point-in-time snapshots for the vendor’s Compute Engine instance and Container Engine services. It’s similar to Amazon Elastic Block Store (EBS), and it offers a choice of either high-throughput HDD, priced at $0.04 per gigabyte per month, or low-latency SSD, priced at $0.17 per gigabyte per month. Both disk types are available in a range of volume sizes from 10 GB to 64 TB.
Features include automatic encryption of data at rest and multireader mounting, whereby many virtual machines can share data on a single Persistent Disk with no degradation in performance.
Local SSDs are low-latency attached storage for Compute Engine instances and are the Google equivalent of AWS Instance Store. They’re particularly useful for applications that benefit from fast read and write speeds. Storage is ephemeral, however, meaning data only persists as long as your instance is running. Nevertheless, your data is still preserved if you restart your machine, but not if you stop it.
Each local SSD is 375 GB in size and charged at $0.218 per gigabyte per month. But if you need more space, you can attach up to eight local SSDs to any machine, giving you up to 3 TB in total.
Google’s object-storage service is the most economical of its general storage options. In line with other object-storage systems, Cloud Storage is highly scalable and handles unstructured data that comes in a variety of formats, such as pictures, text documents and log files.
Cloud Storage is available in four different storage classes, each tailored to specific uses:
- Multi-regional: The vendor’s premium Cloud Storage option, Multi-regional offers the highest level of availability at 99.95%. It’s geo-redundant, which means Google stores your data redundantly across at least two regions in your selected multi-regional location. Similar to Amazon S3 Standard, it’s suitable for storing frequently accessed data, such as website content, interactive workloads and multimedia. It is currently priced at $0.026 per gigabyte per month.
- Regional: Like its Multi-regional counterpart, Regional is designed for frequent data access, but with redundancy restricted to a single region. It’s best suited to local, data-intensive applications that require low end-to-end latency, such as general compute, video transcoding and data analytics. It comes with a slightly lower availability expectation of 99.9%, but also costs slightly less at $0.02 per gigabyte per month.
- Nearline: A lower-price archival storage option similar to Amazon S3 Standard—Infrequent Access, Nearline is best suited to applications that access your objects less than once a month, such as long-tail web content, backups and data stored for occasional analysis. Although availability is lower at 99%, it costs much less at just $0.01 per gigabyte per month. But you pay a charge of $0.01 per gigabyte for accessing your data. Data deleted within 30 days is also subject to minimum 30-day charge.
- Coldline: Google’s new cold-storage service, along the lines of Amazon Glacier, stores data that’s typically accessed less than once a year. It’s the lowest-price option in the Cloud Storage family at a cost of just $0.007 per gigabyte per month. This low price, however, comes at a tradeoff, as the cost of data retrieval is somewhat higher than that for Nearline at $0.05 per gigabyte. Nevertheless, the service still offers users 99% availability, with data access at ultralow millisecond latency on a par with the vendor’s other storage classes. Ideal applications include archiving, source-file backup and disaster recovery.
Cloud Storage also includes built-in redundancy and automatic encryption of data when it travels from your Compute Engine machines to your storage buckets.
Specialist Database Storage
Cloud SQL is a fully managed cloud-based MySQL service whose nearest AWS equivalent is Amazon RDS. Using Cloud SQL, you can do anything that you can with a traditional MySQL database-management system. You connect to it from a Compute Engine instance and query or modify your databases just as you would with MySQL. The only difference is that Google takes care of all database-management tasks for you, such as backups and server updates.
Google’s latest pricing model is structured along the same lines as that of its Compute Engine, consisting of the hourly rate of your database-dedicated machine, your monthly storage cost per gigabyte and network costs.
Cloud SQL is a good solution if you want to move your existing MySQL databases to the cloud and your scaling needs aren’t too great.
Cloud Datastore is a highly scalable and cost-effective NoSQL database as a service that’s similar to Amazon DynamoDB. It is a structured database solution, but based on a loose approach to schema definitions. It can handle databases both large and small, scaling seamlessly and automatically while maintaining performance and relational integrity on a massive scale.
The service handles a diverse range of data types and is typically suited to storing user profiles and product catalogs. Cloud Datastore is also the ideal storage option for applications that start out small but have the potential to grow to a huge scale—you pay by the operation rather than by the hour, and you also get free storage of up 1 GB. Stored data above the 1 GB limit is charged at $0.18 per gigabyte per month. Entity reads cost $0.06 per 100,000, and entity writes cost $0.18 per 100,000.
Also along the lines of Amazon DynamoDB, Cloud Bigtable is another fully managed, scalable NoSQL database service, but it’s geared toward storing, querying, processing and analyzing huge amounts of data.
Offering low latency and high throughput performance, it can handle exceptionally large volumes of read and writes. It’s accessible through the open-source HBase API, making it compatible with big data tools such as Hadoop and Spark.
Prices depend on whether you use SSD or HDD storage, as well as your monthly outbound data transfers and the number of nodes in each cluster. Bigtable supports a minimum cluster size of three nodes, which is enough to handle far more than what a small application needs. The service therefore only comes into its own at terabyte data scale.
Similar to Amazon Redshift, BigQuery is a massively parallel query data store and is therefore more accurate an online analytical-processing (OLAP) tool than a database-storage service. It can run SQL-like queries on terabytes of data in just seconds, making it suitable for big data and business-intelligence applications.
BigQuery is purely for analytical purposes and isn’t designed for use as an operational database. In other words, it’s unsuitable for processing real-time data, but it’s ideal for analyzing recently ingested and historical information.
Standard storage costs $0.02 per gigabyte per month, streaming inserts cost $0.05 per gigabyte and queries cost $5 per terabyte—although your first 1TB of queries each month are free. Loading, copying and exporting data are also free of charge.
A Platform for the Future
Google Cloud Platform is still playing catch-up with AWS and Microsoft Azure, both in enterprise features and in market share. But it’s also a highly competitive offering that’s evolving fast. It puts a strong emphasis on self-service, API-accessible capabilities and cloud-native applications, making it a particularly good choice for new projects born in the cloud.
Although other competitors will fall further and further behind the big two leading cloud players, Google Cloud Platform is built with the future in mind and is almost certainly here to stay.
About the Author
Jonathan Maresky is director of product marketing at Cloudyn, a leading provider of hybrid- and multi-cloud management and optimization solutions for enterprises, CSPs and MSPs. Having more than two decades of technology-industry experience, Jonathan has worked with large multinational technology companies and startups in marketing, technical-account management, professional services and strategic project management, and he has extensive experience in Europe, North America and Asia.