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

On 30.04.2017 16:25, Steve Atkins wrote:
> You can use postgresql for caching, but caches don't require the data
> durability that a database offers, and can be implemented much more
> efficiently.

I for one can understand Thomas' need for a single solution.
Just recently I needed a cache which was supposed to be set up in a
SERIALIZABLE manner as in
https://www.postgresql.org/docs/devel/static/transaction-iso.html#xact-serializable
Available cache mechanisms would have produce erroneous results. So, I
went for PG.

But it's still a cache, isn't it?

> You can use postgresql to provide message queue services and it
> does so reasonably well, particularly when the messages are generated within
> the database. But it's not going to do so as efficiently, or be as easy to
> monitor, to make highly redundant or to scale across a whole datacenter
> as a dedicated message queue service.
>
> You could use postgresql to store binary blobs, but it'd be a horrifically
> inefficient way to do it. (Using postgresql to store the metadata, while
> the content is stored elsewhere, sure).
>
> Use the right tool for the job.

I think it's not as easy as ads and buzz words make us believe it is.

Especially when it comes to reinventing the wheel, I prefer a single
solution. With the better JSON support, PG made NoSQL obsolete. I don't
see why this cannot happen with blob storage and massive scale out. Just
a matter of time, if you ask me.

Regards,
Sven