Volumez block storage for PostgreSQL on AWS, Azure, and Oracle Cloud Infrastructure (OCI) delivers more transactions per second at lower latency than the cloud providers’ own block storage instances.
PostgreSQL is an open source relational database. PostgreSQL, with SQL support, succeeded POSTGRES, which stood for Post-Ingres; Ingres was an earlier relational database. The PostgreSQL database has become widely deployed. Volumez pools cloud block ephemeral storage instances and provides block storage as a shared service to its customers more efficiently and at less cost than the raw AWS, Azure, or OCI services on which it is based. ArchitectingIT is a storage technology analysis service, and analyst Chris Evans has tested Volumez DIaaS (Data Infrastructure-as-a-Service) block storage for PostgreSQL. Evans says DIaaS “abstracts away the implementation details of cloud storage, replacing it with the capability to define application-specific metrics including bandwidth, throughput (IOPS) and latency per storage volume, irrespective of volume capacity and based on dynamic policies set by the administrator (most public cloud storage aligns capacity and IOPS in a linear scale).”
For example, Volumez can separate capacity from performance, typically combined in clouds like AWS and Azure, which means you can over-provision on capacity to get to a performance level.
Evans measured PostgreSQL transactions per second and latency with the pgbench evaluation tool. He ran “pgbench against a set of AWS, Azure and OCI configurations, mapping the results against the underlying cost of deploying that infrastructure.”
There are seven tests in his report:
- AWS EBS io1 storage attached to the database server
- AWS EBS io2 Block Express storage installed on the database server
- Microsoft Azure PV2 SSD storage connected to the database server
- Volumez DIaaS storage on AWS connected to the database server
- Volumez DIaaS storage on Azure connected to the database server
- Volumez DIaaS storage on OCI connected to the database server
- Amazon Aurora storage directly configured as a cluster volume
Evans saw that block storage costs differed with each cloud, OCI being the cheapest and 40 percent less than AWS:
The AWS tests showed that Volumez-PostgreSQL delivered the highest transactions per second (TPS) rating:
He notes: “This data is interesting in that it highlights the relative instability of io1 storage, which significantly underperforms the alternate configurations. This is no surprise, as io1 is a much earlier implementation of SSD-based storage on AWS, compared to io2 Block Express or Volumez DIaaS using NVMe devices.”
The Volumez-PostgreSQL combo also delivered the lowest latency of the alternatives on AWS:
Evans examined the pricing of the various options, saying: “When the pricing is also taken into consideration ($34,000 per month for io2 Express vs $5,000 for the Volumez media nodes, plus 20 percent licensing charge), then the Volumez configuration clearly provides greater value for money.”
Volumez DIaaS also outperformed raw Azure block storage in TPS terms:
And in latency terms as well:
When Evans looked at Volumez and PostgreSQL on OCI he found that “with greater throughput at lower latency, OCI is the best choice for implementing high-performance PostgreSQL applications using Volumez across AWS, Azure and OCI,” as “OCI delivered the lowest latency results compared to AWS and Azure, while also being the cheapest option by a considerable margin.”
In general, “Volumez not only out-performs native block-storage options, but also delivers this capability at a fraction of the cost,” with an approximate “80 percent reduction highlighted for the AWS platform.”
Evans concludes: “Any business using the public cloud at scale, particularly for traditional database applications, should be using Volumez DIaaS as the primary choice for application storage, gaining operationally from flexible storage deployments and being able to gain a level of performance and resiliency that cannot be achieved on the public cloud alone.”
Volumez has published a blog post offering its perspective on the results.
