Re: Can PG replace redis, amqp, s3 in the future?

Am 30.04.2017 um 15:39 schrieb Thomas Delrue:
> On April 30, 2017 1:37:02 PM GMT+02:00, "Thomas Güttler" <guettliml(at)thomas-guettler(dot)de> wrote:
>> Is is possible that PostgreSQL will replace these building blocks in
>> the future?
>>
>> - redis (Caching)
>> - rabbitmq (amqp)
>> - s3 (Blob storage)
>
> These are three very different sets of functionalities, each requiring a different approach. I am curious as to why you are thinking about having a single piece of software that does these three very different things.

I love transactions. If you store data in four systems (three from above plus PG), then you have
better performance if you have high load. But what happens if a transaction fails (rolls back). Then this
can leave the other data sinks in a broken state. Example: a blob in s3 might be updated, but the rollback
in PG does not rollback in s3 ....

And one other benefit if you have one system: Configuration management is easier.

I know that if you have very high load, then you need to optimize.

But in my context the load is far from high. Robust transactions (including
rollback in all related systems) is more important for me.

For the rabbitmq/amqp part I found that the new SKIP LOCKED feature can help:

https://blog.2ndquadrant.com/what-is-select-skip-locked-for-in-postgresql-9-5/

Leaving two things open:

- blob storage
- redis/caching

>> One question is "is it possible?", then next "is it feasible?"
>
> Possible? Sure: p != 0
> Probable? No
> Desirable? No
>
>> I think it would be great if I could use PG only and if I could
>> avoid the other types of servers.
>
> When you're holding a hammer, everything looks like a nail. But hammering screws doesn't get you very far. Sometimes you need a screwdriver and on other days a glue gun...

Yes, you are right.

Regards,
Thomas Güttler

--
I am looking for feedback for my personal programming guidelines:
https://github.com/guettli/programming-guidelines