Re: Interesting read on SCM upending software and hardware architecture

W dniu 18.01.2016, pon o godzinie 18∶55 -0600, użytkownik Jim Nasby
napisał:
[ cut ]
> 
> My original article doesn't talk about SSDs; it's talking about 
> non-volatile memory architectures (quoted extract below). Fusion IO
> is 
> an example of this, and if NVDIMMs become available we'll see even 
> faster non-volatile performance.
>
> To me, the most interesting point the article makes is that systems
> now 
> need much better support for multiple classes of NV storage. I agree 
> with your point that spinning rust is here to stay for a long time, 
> simply because it's cheap as heck. So systems need to become much
> better 
> at moving data between different layers of NV storage so that you're 
> getting the biggest bang for the buck. That will remain critical as
> long 
> as SCM's remain 25x more expensive than rust.
>

I guess PostgreSQL is getting ready for such a world.
Parallel sequential scan, while not useful for spinning drives,
should shine with hardware describe in that article.

Add some tuning of effective_io_concurrency and we might
have some gains even without new storage layer.
Of course ability to change storage subsystem might
help with experimentation, but even now (OK, when 9.6 is out)
we might use increased IO concurrency.

-- 
Tomasz Rybak GPG/PGP key ID: 2AD5 9860
Fingerprint A481 824E 7DD3 9C0E C40A 488E C654 FB33 2AD5 9860
http://member.acm.org/~tomaszrybak