Customers fret about downtime with hyperscalers' PostgreSQL services
(2025/07/22)
- Reference: 1753172887
- News link: https://www.theregister.co.uk/2025/07/22/postgresql_hyperscalers_gartner/
- Source link:
Analysis Recent research suggests customers are concerned about the uptime reliability of hyperscalers' PostgreSQL instances, giving smaller alternative vendors an opening to fill the gap.
Research firm Foundry [1]found that among users of common cloud services for PostgreSQL, 82 percent expressed concern about cloud region failures, while 21 percent experienced such failures in the past year.
In the first six months of 2025, database popularity ranking service DB-Engines found PostgreSQL the biggest climber, with a ranking increase of over 13 points. It is fourth overall, behind Oracle, MySQL, and Microsoft SQL Server. In [2]2023 , the open source database became the most popular database for developers, likely because services from the leading three cloud vendors made it easily available to developers.
[3]
But last week's Foundry survey of 212 IT decision-makers across enterprises and SaaS businesses found that 91 percent of organizations currently using PostgreSQL demand no more than four minutes of downtime per month, or around 99.99 percent uptime, while 24 percent aim for less than 30 seconds.
[4]
[5]
The study was commissioned by distributed PostgreSQL vendor pgEdge. Gartner's Adam Ronthal, veep and analyst, said was unclear whether there is a genuine problem with PostgreSQL service reliability SLAs yet. “Certainly, inbound calls to Gartner’s inquiry service does not clearly identify this as an issue at this point in time.”
However, he called on cloud service providers to track and document uptime and reliability to defuse the “rumors” and identify if there is genuinely a problem with specific services.
[6]
Microsoft offers Azure Database for PostgreSQL as a fully managed service with single server, flexible server, and hyperscale options. Both AWS and Google offer basic PostgreSQL services as well as Aurora and AlloyDB as options more tightly integrated with the underlying cloud infrastructure.
Bring on the alternatives
Any users finding that these services do not meet their uptime needs have a range of alternative PostgreSQL and PostgreSQL-compatible services to choose from, with distributed back ends to enhance reliability. Examples include [7]CockroachDB and [8]YugabyteDB .
Meanwhile, [9]PlanetScale , which has made its name with a distributed MySQL service based on YouTube-developed distributed relational database Vitess, launched a PostgreSQL service earlier this month.
Speaking to The Register , CEO Sam Lambert said that rather than compromise compatibility, the PlanetScale service is built on open source PostgreSQL and uses the proprietary operator the company developed for MySQL/Vitess.
“The operator is the piece of magic that kind of is what ensures PlanetScale has such strong uptime and reliability: it's what manages all of the nodes, the state and the topology. It was really just about getting PostgreSQL fitted into that,” he said.
[10]
The current service is not sharded, but a sharded service is in development with customers, Lambert said. It said it would be “test equivalent” for PostgreSQL, but not 100 percent compatible. “If anyone wants to use that, they should migrate to the unsharded service, and we’ll work on getting people to the sharded version from there.”
He said YugabyteDB and CockroachDB were “impressive” distributed database services, but claims they do not offer the same degree of PostgreSQL compatibility. “They're not even that compatible,” he said.
For example, the [11]PostgreSQL Compatibility Index scores CockroachDB at about 40 percent, while Yugabyte achieves about 85 percent compatibility.
Lambert said PlanetScale's current PostgreSQL service was 100 percent compatible while the sharded service would likely be 99 percent compatible when it arrives.
Convex, which offers an open-source reactive database for web app developers, has already started to move from AWS Aurora to PlanetScale. In a blog post the company said: “For years, we've operated Convex on top of AWS Aurora. The experience has been okay. Aurora does what it says on the tin: it's reliable, it scales reasonably well, and it integrates nicely with the rest of AWS. But ‘okay’ isn't where we want to stop.”
On the select projects already using the new service, “performance gains are incredible”, the [12]company said .
Another company looking to take advantage of the questions about hyperscalers’ PostgreSQL service reliability, and the compatibility of the distributed systems, is pgEdge, which promises multi-master distributed database built on open-source PostgreSQL, allowing users to have multiple master databases spread across different locations.
Speaking to El Reg co-founder and CEO Phillip Merrick said the fully PostgreSQL-compatible service offers “pretty extreme levels of high availability”.
As a by-product, the service also helps address the lag problem for application users needing data from a database in different parts of the world. “For applications that are being used globally, when you're accessing websites or apps that are hosted entirely in the US, from the UK and Europe, you're going to see a bit of a lag, just because the speed of light is a thing, and it takes a while for those bits to go back and forth across the ocean.”
Typical customers are large organizations, with a number in financial services, he said. “Most of our customers are financial services, government, other large enterprises, where the application simply can't go down, so there's almost zero tolerance for any downtime, and as a result, they need extreme levels of high availability, coupled with the need to be able to failover for disaster recovery protection from one geographical region to another,” he said.
Gartner’s Ronthal said distributed PosgreSQL services can insulate from the failure of a single cloud region. “The organizations that tend to need this degree of fault tolerance represent a relatively small segment of the market. The tradeoff is usually around core features. You get built in distributed processing and transactional consistency, but maybe give up some other advanced DBMS features,” he said.
[13]Users of PostgreSQL in the cloud say the uptime just ain't up to it
[14]Atlassian migrated 4 million Postgres databases to shrink AWS bill
[15]Coming to PostgreSQL: On-disk database encryption
[16]French city of Lyon ditching Microsoft for open source office and collab tools
Such distributed services might be an option for users of hyperscalers’ PostgreSQL services if they implement the same SQL API, Ronthal said. “The real question is whether these offerings provide compelling price/performance, and whether RDS users need that degree of high availability. Some use cases do, some don’t. RDS is a transition stepping stone to more capable services – whether they are tightly integrated with the underlying cloud infrastructure like Aurora or AlloyDB, or 3rd parties that offer something different like the distributed SQL Database vendors.”
In response to the survey, and outstanding questions over the reliability of PostgreSQL services, AWS said its PostgreSQL-compatible Amazon Aurora DSQL was designed for up to 99.999 percent availability while Amazon Aurora for PostgreSQL offers a 99.99 percent uptime SLA and global replication with cross-Region disaster recovery in less than 1 minute. The spokesperson said AWS provided documentation to explain what customers should expect in terms of downtime during upgrades.
Google and Microsoft were also contacted about the issues raised in this article. ®
Get our [17]Tech Resources
[1] https://www.pgedge.com/blog/postgresql-in-mission-critical-environments-survey-results
[2] https://devclass.com/2023/06/13/postgresql-now-top-developer-choice-ahead-of-mysql-according-to-massive-new-survey/
[3] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/databases&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2aH9hNIsJymEIiDBgnz4okwAAAg0&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0
[4] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/databases&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44aH9hNIsJymEIiDBgnz4okwAAAg0&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[5] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/databases&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33aH9hNIsJymEIiDBgnz4okwAAAg0&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[6] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/databases&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44aH9hNIsJymEIiDBgnz4okwAAAg0&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[7] https://www.theregister.com/2024/09/27/cockroach_labs_ceo/
[8] https://www.theregister.com/2022/06/28/yugabytedb_215/
[9] https://www.theregister.com/2022/03/24/planetscale_undo_button/
[10] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/databases&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33aH9hNIsJymEIiDBgnz4okwAAAg0&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[11] https://pgscorecard.com/
[12] https://news.convex.dev/powered-by-planetscale-for-postgres/
[13] https://www.theregister.com/2025/07/11/cloud_postgresql_uptime/
[14] https://www.theregister.com/2025/07/07/asia_tech_news_in_brief/
[15] https://www.theregister.com/2025/07/02/postgresql_ondisk_database_encryption/
[16] https://www.theregister.com/2025/06/26/lyon_leaving_microsoft/
[17] https://whitepapers.theregister.com/
Research firm Foundry [1]found that among users of common cloud services for PostgreSQL, 82 percent expressed concern about cloud region failures, while 21 percent experienced such failures in the past year.
In the first six months of 2025, database popularity ranking service DB-Engines found PostgreSQL the biggest climber, with a ranking increase of over 13 points. It is fourth overall, behind Oracle, MySQL, and Microsoft SQL Server. In [2]2023 , the open source database became the most popular database for developers, likely because services from the leading three cloud vendors made it easily available to developers.
[3]
But last week's Foundry survey of 212 IT decision-makers across enterprises and SaaS businesses found that 91 percent of organizations currently using PostgreSQL demand no more than four minutes of downtime per month, or around 99.99 percent uptime, while 24 percent aim for less than 30 seconds.
[4]
[5]
The study was commissioned by distributed PostgreSQL vendor pgEdge. Gartner's Adam Ronthal, veep and analyst, said was unclear whether there is a genuine problem with PostgreSQL service reliability SLAs yet. “Certainly, inbound calls to Gartner’s inquiry service does not clearly identify this as an issue at this point in time.”
However, he called on cloud service providers to track and document uptime and reliability to defuse the “rumors” and identify if there is genuinely a problem with specific services.
[6]
Microsoft offers Azure Database for PostgreSQL as a fully managed service with single server, flexible server, and hyperscale options. Both AWS and Google offer basic PostgreSQL services as well as Aurora and AlloyDB as options more tightly integrated with the underlying cloud infrastructure.
Bring on the alternatives
Any users finding that these services do not meet their uptime needs have a range of alternative PostgreSQL and PostgreSQL-compatible services to choose from, with distributed back ends to enhance reliability. Examples include [7]CockroachDB and [8]YugabyteDB .
Meanwhile, [9]PlanetScale , which has made its name with a distributed MySQL service based on YouTube-developed distributed relational database Vitess, launched a PostgreSQL service earlier this month.
Speaking to The Register , CEO Sam Lambert said that rather than compromise compatibility, the PlanetScale service is built on open source PostgreSQL and uses the proprietary operator the company developed for MySQL/Vitess.
“The operator is the piece of magic that kind of is what ensures PlanetScale has such strong uptime and reliability: it's what manages all of the nodes, the state and the topology. It was really just about getting PostgreSQL fitted into that,” he said.
[10]
The current service is not sharded, but a sharded service is in development with customers, Lambert said. It said it would be “test equivalent” for PostgreSQL, but not 100 percent compatible. “If anyone wants to use that, they should migrate to the unsharded service, and we’ll work on getting people to the sharded version from there.”
He said YugabyteDB and CockroachDB were “impressive” distributed database services, but claims they do not offer the same degree of PostgreSQL compatibility. “They're not even that compatible,” he said.
For example, the [11]PostgreSQL Compatibility Index scores CockroachDB at about 40 percent, while Yugabyte achieves about 85 percent compatibility.
Lambert said PlanetScale's current PostgreSQL service was 100 percent compatible while the sharded service would likely be 99 percent compatible when it arrives.
Convex, which offers an open-source reactive database for web app developers, has already started to move from AWS Aurora to PlanetScale. In a blog post the company said: “For years, we've operated Convex on top of AWS Aurora. The experience has been okay. Aurora does what it says on the tin: it's reliable, it scales reasonably well, and it integrates nicely with the rest of AWS. But ‘okay’ isn't where we want to stop.”
On the select projects already using the new service, “performance gains are incredible”, the [12]company said .
Another company looking to take advantage of the questions about hyperscalers’ PostgreSQL service reliability, and the compatibility of the distributed systems, is pgEdge, which promises multi-master distributed database built on open-source PostgreSQL, allowing users to have multiple master databases spread across different locations.
Speaking to El Reg co-founder and CEO Phillip Merrick said the fully PostgreSQL-compatible service offers “pretty extreme levels of high availability”.
As a by-product, the service also helps address the lag problem for application users needing data from a database in different parts of the world. “For applications that are being used globally, when you're accessing websites or apps that are hosted entirely in the US, from the UK and Europe, you're going to see a bit of a lag, just because the speed of light is a thing, and it takes a while for those bits to go back and forth across the ocean.”
Typical customers are large organizations, with a number in financial services, he said. “Most of our customers are financial services, government, other large enterprises, where the application simply can't go down, so there's almost zero tolerance for any downtime, and as a result, they need extreme levels of high availability, coupled with the need to be able to failover for disaster recovery protection from one geographical region to another,” he said.
Gartner’s Ronthal said distributed PosgreSQL services can insulate from the failure of a single cloud region. “The organizations that tend to need this degree of fault tolerance represent a relatively small segment of the market. The tradeoff is usually around core features. You get built in distributed processing and transactional consistency, but maybe give up some other advanced DBMS features,” he said.
[13]Users of PostgreSQL in the cloud say the uptime just ain't up to it
[14]Atlassian migrated 4 million Postgres databases to shrink AWS bill
[15]Coming to PostgreSQL: On-disk database encryption
[16]French city of Lyon ditching Microsoft for open source office and collab tools
Such distributed services might be an option for users of hyperscalers’ PostgreSQL services if they implement the same SQL API, Ronthal said. “The real question is whether these offerings provide compelling price/performance, and whether RDS users need that degree of high availability. Some use cases do, some don’t. RDS is a transition stepping stone to more capable services – whether they are tightly integrated with the underlying cloud infrastructure like Aurora or AlloyDB, or 3rd parties that offer something different like the distributed SQL Database vendors.”
In response to the survey, and outstanding questions over the reliability of PostgreSQL services, AWS said its PostgreSQL-compatible Amazon Aurora DSQL was designed for up to 99.999 percent availability while Amazon Aurora for PostgreSQL offers a 99.99 percent uptime SLA and global replication with cross-Region disaster recovery in less than 1 minute. The spokesperson said AWS provided documentation to explain what customers should expect in terms of downtime during upgrades.
Google and Microsoft were also contacted about the issues raised in this article. ®
Get our [17]Tech Resources
[1] https://www.pgedge.com/blog/postgresql-in-mission-critical-environments-survey-results
[2] https://devclass.com/2023/06/13/postgresql-now-top-developer-choice-ahead-of-mysql-according-to-massive-new-survey/
[3] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/databases&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2aH9hNIsJymEIiDBgnz4okwAAAg0&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0
[4] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/databases&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44aH9hNIsJymEIiDBgnz4okwAAAg0&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[5] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/databases&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33aH9hNIsJymEIiDBgnz4okwAAAg0&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[6] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/databases&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44aH9hNIsJymEIiDBgnz4okwAAAg0&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[7] https://www.theregister.com/2024/09/27/cockroach_labs_ceo/
[8] https://www.theregister.com/2022/06/28/yugabytedb_215/
[9] https://www.theregister.com/2022/03/24/planetscale_undo_button/
[10] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/databases&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33aH9hNIsJymEIiDBgnz4okwAAAg0&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[11] https://pgscorecard.com/
[12] https://news.convex.dev/powered-by-planetscale-for-postgres/
[13] https://www.theregister.com/2025/07/11/cloud_postgresql_uptime/
[14] https://www.theregister.com/2025/07/07/asia_tech_news_in_brief/
[15] https://www.theregister.com/2025/07/02/postgresql_ondisk_database_encryption/
[16] https://www.theregister.com/2025/06/26/lyon_leaving_microsoft/
[17] https://whitepapers.theregister.com/