Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.databases.postgresql > #314

Re: MySQL

Show all headers | View raw

Hi Mladen,

On 2/16/2012 3:05 PM, Mladen Gogala wrote:
> It looks that the other open source database is making great leaps in
> performance:
>
> http://www.mysql.com/why-mysql/white-papers/mysql-cluster-7.2-ga.html
>
> In addition to better partitioning and parallelism, MySQL also has hints.
> That may motivate PostgreSQL developers to make some of the features
> available in free PostgreSQL implementation. So far, the features like
> the ones listed above were only available in commercial versions of
> Postgres, like VoltDB, Vertica or EnterpriseDB. With this new version,
> MySQL will run circles around Postgres, which may make Postgres much less
> interesting for the people trying to get off Oracle.

As with any market, you pick the product that best fits your
*particular* selection criteria.  Some mix of features, performance,
cost, reliability, etc.  The closer to "mainstream" that your choices
are, the more selection you will *tend* to have available.

That doesn't mean suppliers won't exist for niche markets or
for markets that they perceive to be headed in a different
direction, etc.

(e.g., TI made processors ~25 years ago that had no internal
"registers" [1].  They assumed memory would get faster AND
the constraint on some small, fixed number of internal
registers would be a wise one to lift.  Unfortunately, for TI,
technology and The Industry went in a different direction!)

We had hoped PG would be a good fit for some products, here.
(some felt MySQL would be The Right Choice).  We knew it was
missing many features that we sought.  But, hoped those features
would come along, with time.  We keep Oracle docs on hand to
make sure we don't stray too far afield with particular PG/MySQL
features:  "is there another, less PG/MySQL-specific, way of
doing this?"

I am slowly (and reluctantly) coming to the conclusion that
we will probably end up as Oracle customers -- simply because
some of the "must have" features will be too hard to back-port
(reliably!) into other implementations if they differ too much
from the "design philosophy" of those original implementations
(e.g., read only media support is one such issue.  others are
even more difficult to get into the pie after its been baked!)

A "proprietary" approach may prove to be the most appropriate
solution (<huge_frown>).

--------------------
[1]  This is a minor inaccuracy; the workspace pointer, etc.
resided in the CPU.

Back to comp.databases.postgresql | Previous | Next — Previous in thread | Next in thread | Find similar

Thread

MySQL Mladen Gogala <gogala.mladen@gmail.com> - 2012-02-16 22:05 +0000
  Re: MySQL Harry Tuttle <OTPXDAJCSJVU@spammotel.com> - 2012-02-17 15:49 +0100
    Re: MySQL Mladen Gogala <gogala.mladen@gmail.com> - 2012-02-18 03:06 +0000
      Re: MySQL Hans Castorp <hans.castorp@xyz.de> - 2012-02-18 07:56 +0100
        Re: MySQL Mladen Gogala <gogala.mladen@gmail.com> - 2012-02-18 18:40 +0000
  Re: MySQL Don Y <this@isnotme.com> - 2012-02-18 12:17 -0700
    Re: MySQL Hans Castorp <hans.castorp@xyz.de> - 2012-02-18 22:07 +0100
      Re: MySQL Don Y <this@isnotme.com> - 2012-02-18 15:13 -0700
      Re: MySQL Mladen Gogala <gogala.mladen@gmail.com> - 2012-02-18 22:29 +0000

csiph-web