Google Cloud Platform: Key storage options in the Google Cloud | JP
Google’s first step into the cloud was App Engine, launched in 2008. Google Cloud Storage followed later in 2010. Since then, the company has steadily expanded to what is now known as the Google Cloud Platform (GCP) and consolidated its position as one of the three, global “hyperscale” cloud storage providers.
Google Cloud is seen as the most technical of the three hyperscalers, and analysts rank it third – after Amazon Web Services (AWS) and Microsoft Azure – in market share. Google continues to expand its storage offerings, and has a comprehensive suite of technologies. Some, such as Cloud Storage, are pure-play storage offerings. Others, such as BigQuery, go after specific markets, with storage as part of the service.
Google Cloud: storage options
Google has three main cloud storage products: Cloud Storage, Persistent Disk and Filestore. Pricing is based on storage volume, storage class, frequency of access, performance required and, in some cases, network and access or egress charges. Google also provides a range of locations, including single-, dual- and multi-zone geographies.
Cloud Storage
Google’s original offering, Cloud Storage is an object storage system that supports four storage tiers. Pricing varies depending on the tier, and also the geographical regions used.
Standard storage is for “hot” data that needs immediate access, with a service-level agreement (SLA) of 99.9% for a single-region store. Nearline storage is designed to store data for at least 30 days and accessed at least once a month.
Nearline is suggested for backups, media files and archives. It has a slightly lower SLA, at 99%. Google, though, recommends Coldline and Archive storage for longer-term data retention. Coldline and Archive have the same SLAs as Nearline.
For a single region (London, Europe-West2), Cloud Storage costs $0.023 per gigabyte (GB) per month for Standard, $0.013 for Nearline, $0.007 for Coldline and $0.0025 for Archive.
There are also operational charges, based on the namespaces used. Each tier other than Standard has a retrieval charge ($0.01/GB for Nearline, $0.02/GB for Coldline and $0.05/GB for Archive).
Persistent Disk
Persistent Disk is GCP’s offering targeted at virtualised environments. Persistent Disk is provided over conventional hard disk drives (HDDs) or solid-state drives (SSDs), with a range of performance tiers based on input/output operations per second (IOPS) and throughput. Persistent Disk also offers snapshots, encryption and high availability through cross-region replication.
Persistent Disk pricing varies according to the type of virtual machine (VM) the customer runs. Google charges usage fees for “premium” disk images. These include Red Hat Enterprise Linux, including RHEL for SAP, SUSE Linux Enterprise Server (and SLES for SAP), Ubuntu Pro Windows Server, and SQL Server.
The cost also depends on the server itself. A Google f1-micro VM, for example, costs $0.02 an hour, and an n1-standard-8 costs $0.11 an hour. The final price consists of machine and image fees. These can be complex, and IT teams should refer to Google’s pricing calculators.
Filestore
Filestore is Google’s managed file storage or cloud NAS service. Filestore uses the Network File System (v3 and v4.1) and offers hard disks or SSDs depending on user performance requirements.
Filestore can connect to on-premise machines, Google Compute Engine VMs, Google Kubernetes Engine clusters, external data stores and Cloud Run services. Tiers comprise the Basic tier, Regional tier, Enterprise tier and Zonal tier, with the latter aimed at workloads including high-performance computing (HPC), advanced artificial intelligence (AI) and large datasets. Google also offers backups and snapshots within Filestore.
Pricing ranges from $0.19 per gibibyte (GiB) per month (London) for Basic HDD, $0.36 for Basic SSD, $0.30 for Zonal and High Scale storage and $0.55 for Regional and Enterprise tiers. Pricing is based on provisioned capacity, not actual use.
BigQuery, Cloud SQL Firestore and AlloyDB
Google Cloud Platform also includes storage with its database and analytics products.
CloudSQL allows users to tailor storage according to their needs when running relational databases in the cloud. Storage pricing for CloudSQL ranges from $0.108/GB per month for HDD storage to $0.204/GB per month for SSDs (both based on London). High-availability storage costs more.
Firestore is a fully managed NoSQL document database which, Google says, includes automatic scaling as well as offline capabilities. AlloyDB, for its part, is Google’s PostgreSQL-compatible cloud database. Storage for AlloyDB includes regional cluster storage, starting at $0.0004109/GB per hour, and backup storage at $0.000137/GB per hour.
BigQuery works in a slightly different way, as storage is integrated with the BigQuery stack. Google emphasises that BigQuery users do not have to provision or reserve storage. Instead, the platform automatically allocates storage as the user uploads data.
This means there is no cost for provisioned but unused storage, although BigQuery does charge separately for compute and storage. BigQuery also automatically stores data across availability zones, and the hyperscaler says the system is designed for 11 nines availability.
Costs range from $0.02/GiB per month for Active logical storage and $0.044/GiB per month, down to $0.01/GiB per month for long-term logical storage. The first 10GiB is free each month in each case.