Обсуждение: PostgreSQL as a Service
Hello everyone! tl;dr: How well is PostgreSQL positioned to serve as the database of choice for a DBaaS operator? Specifically, how much open source is (may be) missing? ---- Im un-lurking hoping to learn more about PostgreSQL in DBaaS land. You may have seen this announcement. https://blog.yugabyte.com/why-we-changed-yugabyte-db-licensing-to-100-open-source/ YugaByte bills itself as a PostgreSQL compatible database (yay to at least the intent) but most importantly, it decided to single-license its database under a permissive license, including "the enterprise features" that frequently are held back by single-vendor open source firms who want to earn a RoI for their VC investment. The interesting part (and why I'm posting it here) is the following staging of functionality implied in that post. 1. Core database (permissively licensed) 2. Enterprise features (permissively licensed) 3. DBaaS features (trial license, commercial, no open source) 4. Managed by YugaByte (commercial) Point 3. suggests that they want to make money from self-managed DBaaS, but in the post they also write they really only expect significant income from 4, i.e. YugaByte (the database) managed by YugaByte (the company). Where is PostgreSQL in relation to this? 1. PostgreSQL itself is certainly 1 above, the core database. 2. PostgreSQL permissive license allows commercial offerings to build and not share enterprise features (and I'm sure some companies are holding back). However, PostgreSQL is true community open source so whatever enterprise features become relevant, they'll eventually be commoditized and out in the open. Is there a lot that is missing? And that some companies have but are not contributing? 3. So, PostgreSQL as-a-service. There are several companies (plenty?) who service PostgreSQL. I wonder how this is being shared back? I don't have a clear picture here, my impression is that the software to run these potentially large farms is proprietary? Or, that operators would argue, this is all configuration and shell scripts and not really shareable open source? One aspect related to as-a-service is scaling out, i.e. not just having many small customers, but also serving large customers in the cloud. I looked around for scaling out solutions. There used to be CitusData (not any longer it seems), there is PostgresXL which seems to be moving slowly. Is that it? 4. Managed DBaaS is not relevant here but always a commercial offering. So, back to my main question above. If I wanted to run a DBaaS shop with only PostgreSQL open source, how far away from being able to compete with AWS or Azure (or YugaByte for that matter) would I be? Thanks for any thoughts and opinions! Dirk -- Website: http://dirkriehle.com - Twitter: @dirkriehle Ph (DE): +49-157-8153-4150 - Ph (US): +1-650-450-8550
On 18/7/19 5:23 μ.μ., Dirk Riehle wrote: > Hello everyone! > > tl;dr: How well is PostgreSQL positioned to serve as the database of choice for a DBaaS operator? Specifically, how muchopen source is (may be) missing? > > ---- > > Im un-lurking hoping to learn more about PostgreSQL in DBaaS land. > > You may have seen this announcement. > > https://blog.yugabyte.com/why-we-changed-yugabyte-db-licensing-to-100-open-source/ > > YugaByte bills itself as a PostgreSQL compatible database (yay to at least the intent) but most importantly, it decidedto single-license its database under a permissive license, including "the > enterprise features" that frequently are held back by single-vendor open source firms who want to earn a RoI for theirVC investment. > > The interesting part (and why I'm posting it here) is the following staging of functionality implied in that post. > > 1. Core database (permissively licensed) > 2. Enterprise features (permissively licensed) > 3. DBaaS features (trial license, commercial, no open source) > 4. Managed by YugaByte (commercial) > > Point 3. suggests that they want to make money from self-managed DBaaS, but in the post they also write they really onlyexpect significant income from 4, i.e. YugaByte (the database) managed by > YugaByte (the company). > > Where is PostgreSQL in relation to this? > > 1. PostgreSQL itself is certainly 1 above, the core database. > > 2. PostgreSQL permissive license allows commercial offerings to build and not share enterprise features (and I'm sure somecompanies are holding back). However, PostgreSQL is true community open > source so whatever enterprise features become relevant, they'll eventually be commoditized and out in the open. Is therea lot that is missing? And that some companies have but are not contributing? > > 3. So, PostgreSQL as-a-service. There are several companies (plenty?) who service PostgreSQL. I wonder how this is beingshared back? I don't have a clear picture here, my impression is that the > software to run these potentially large farms is proprietary? Or, that operators would argue, this is all configurationand shell scripts and not really shareable open source? > > One aspect related to as-a-service is scaling out, i.e. not just having many small customers, but also serving large customersin the cloud. I looked around for scaling out solutions. There used to > be CitusData (not any longer it seems), there is PostgresXL which seems to be moving slowly. Is that it? There is also the Bidirectional Replication project (BDR). Also Why the comment about CitusData? I would the guess the oppositeis true. > > 4. Managed DBaaS is not relevant here but always a commercial offering. > > So, back to my main question above. If I wanted to run a DBaaS shop with only PostgreSQL open source, how far away frombeing able to compete with AWS or Azure (or YugaByte for that matter) would I be? > > Thanks for any thoughts and opinions! Dirk > -- Achilleas Mantzios IT DEV Lead IT DEPT Dynacom Tankers Mgmt
On 7/18/19 7:23 AM, Dirk Riehle wrote: > Hello everyone! > > tl;dr: How well is PostgreSQL positioned to serve as the database of > choice for a DBaaS operator? Specifically, how much open source is (may > be) missing? > > ---- > > Im un-lurking hoping to learn more about PostgreSQL in DBaaS land. > > You may have seen this announcement. > > https://blog.yugabyte.com/why-we-changed-yugabyte-db-licensing-to-100-open-source/ > > > YugaByte bills itself as a PostgreSQL compatible database (yay to at > least the intent) but most importantly, it decided to single-license its > database under a permissive license, including "the enterprise features" > that frequently are held back by single-vendor open source firms who > want to earn a RoI for their VC investment. > > The interesting part (and why I'm posting it here) is the following > staging of functionality implied in that post. > > 1. Core database (permissively licensed) > 2. Enterprise features (permissively licensed) > 3. DBaaS features (trial license, commercial, no open source) > 4. Managed by YugaByte (commercial) > > Point 3. suggests that they want to make money from self-managed DBaaS, > but in the post they also write they really only expect significant > income from 4, i.e. YugaByte (the database) managed by YugaByte (the > company). > > Where is PostgreSQL in relation to this? > > 1. PostgreSQL itself is certainly 1 above, the core database. > > 2. PostgreSQL permissive license allows commercial offerings to build > and not share enterprise features (and I'm sure some companies are > holding back). However, PostgreSQL is true community open source so > whatever enterprise features become relevant, they'll eventually be > commoditized and out in the open. Is there a lot that is missing? And > that some companies have but are not contributing? > > 3. So, PostgreSQL as-a-service. There are several companies (plenty?) > who service PostgreSQL. I wonder how this is being shared back? I don't > have a clear picture here, my impression is that the software to run > these potentially large farms is proprietary? Or, that operators would > argue, this is all configuration and shell scripts and not really > shareable open source? > > One aspect related to as-a-service is scaling out, i.e. not just having > many small customers, but also serving large customers in the cloud. I > looked around for scaling out solutions. There used to be CitusData (not > any longer it seems), there is PostgresXL which seems to be moving > slowly. Is that it? > > 4. Managed DBaaS is not relevant here but always a commercial offering. > > So, back to my main question above. If I wanted to run a DBaaS shop with > only PostgreSQL open source, how far away from being able to compete > with AWS or Azure (or YugaByte for that matter) would I be? The difference in resources available. The pull of DBaaS as I see it is the being able to spin up db's as needed on a scale needed from one or more locations. All with a unified management fronted/API. Being competitive means being able to match that. > > Thanks for any thoughts and opinions! Dirk > -- Adrian Klaver adrian.klaver@aklaver.com
Thanks for the pointer to BDR!
As to CitusData, you are right. The Microsoft acquisition does not seem to have led to absorption. Rather, it remains an open core play. This also means its DBaaS layer is not open source.
Cheers, Dirk
On Thu, Jul 18, 2019, 16:50 Achilleas Mantzios <achill@matrix.gatewaynet.com> wrote:
On 18/7/19 5:23 μ.μ., Dirk Riehle wrote:
> Hello everyone!
>
> tl;dr: How well is PostgreSQL positioned to serve as the database of choice for a DBaaS operator? Specifically, how much open source is (may be) missing?
>
> ----
>
> Im un-lurking hoping to learn more about PostgreSQL in DBaaS land.
>
> You may have seen this announcement.
>
> https://blog.yugabyte.com/why-we-changed-yugabyte-db-licensing-to-100-open-source/
>
> YugaByte bills itself as a PostgreSQL compatible database (yay to at least the intent) but most importantly, it decided to single-license its database under a permissive license, including "the
> enterprise features" that frequently are held back by single-vendor open source firms who want to earn a RoI for their VC investment.
>
> The interesting part (and why I'm posting it here) is the following staging of functionality implied in that post.
>
> 1. Core database (permissively licensed)
> 2. Enterprise features (permissively licensed)
> 3. DBaaS features (trial license, commercial, no open source)
> 4. Managed by YugaByte (commercial)
>
> Point 3. suggests that they want to make money from self-managed DBaaS, but in the post they also write they really only expect significant income from 4, i.e. YugaByte (the database) managed by
> YugaByte (the company).
>
> Where is PostgreSQL in relation to this?
>
> 1. PostgreSQL itself is certainly 1 above, the core database.
>
> 2. PostgreSQL permissive license allows commercial offerings to build and not share enterprise features (and I'm sure some companies are holding back). However, PostgreSQL is true community open
> source so whatever enterprise features become relevant, they'll eventually be commoditized and out in the open. Is there a lot that is missing? And that some companies have but are not contributing?
>
> 3. So, PostgreSQL as-a-service. There are several companies (plenty?) who service PostgreSQL. I wonder how this is being shared back? I don't have a clear picture here, my impression is that the
> software to run these potentially large farms is proprietary? Or, that operators would argue, this is all configuration and shell scripts and not really shareable open source?
>
> One aspect related to as-a-service is scaling out, i.e. not just having many small customers, but also serving large customers in the cloud. I looked around for scaling out solutions. There used to
> be CitusData (not any longer it seems), there is PostgresXL which seems to be moving slowly. Is that it?
There is also the Bidirectional Replication project (BDR). Also Why the comment about CitusData? I would the guess the opposite is true.
>
> 4. Managed DBaaS is not relevant here but always a commercial offering.
>
> So, back to my main question above. If I wanted to run a DBaaS shop with only PostgreSQL open source, how far away from being able to compete with AWS or Azure (or YugaByte for that matter) would I be?
>
> Thanks for any thoughts and opinions! Dirk
>
--
Achilleas Mantzios
IT DEV Lead
IT DEPT
Dynacom Tankers Mgmt
On 7/18/19 9:06 AM, Dirk Riehle wrote: Please reply to list also. Ccing list. > > > On Thu, Jul 18, 2019, 16:56 Adrian Klaver <adrian.klaver@aklaver.com > <mailto:adrian.klaver@aklaver.com>> wrote: > > > > So, back to my main question above. If I wanted to run a DBaaS > shop with > > only PostgreSQL open source, how far away from being able to compete > > with AWS or Azure (or YugaByte for that matter) would I be? > > The difference in resources available. The pull of DBaaS as I see it is > the being able to spin up db's as needed on a scale needed from one or > more locations. All with a unified management fronted/API. Being > competitive means being able to match that. > > > Yes that's the point. I'm not aware of an open source DBaaS software > layer for PostgreSQL. Are there any attempts? I missed that, I thought you where referring only to the Postgres database component only. So you want the management infrastructure to be Open Source also. I am not aware of any, but I don't work in the DBaaS field so take my observation in that light. > > Cheers, Dirk > > > > > > > Thanks for any thoughts and opinions! Dirk > > > > > -- > Adrian Klaver > adrian.klaver@aklaver.com <mailto:adrian.klaver@aklaver.com> > -- Adrian Klaver adrian.klaver@aklaver.com