On Mon, 2004-12-06 at 06:53, Chad Perrin wrote:
> That all makes sense, but the matter of clustering can make a
> significant difference when making decisions about which to use for a
> given task. Contrasted with that, licensing issues, scalability, and
> the more-complete SQL implementation in PostreSQL can be quite
> non-trivial for certain tasks as well.
The term "clustering" is rather broad. Load balancing, replication,
virtualization, failover, etc... = lots of factors. I've handled
replication with both MySQL and PostgreSQL in a number of ways. I could
support either for different applications, or even the same.
But let's get past the theoretical, let's talk a _real_ implementation.
Let's say I had a web site load balanced over running over 4 web
servers. Like most web applications, I'm largely hitting the DB
read-only. Now what could I do?
I could have a dedicated MySQL server. The 4 web servers then replicate
a read-only copy locally. Any commits are then made back to the single,
dedicated MySQL server. Given the fact that I'm largely reading, the
commits to that dedicated MySQL server are still going to be far less
than the hits on the web servers and their local, read-only replications
of the DB.
Or I could use PostgreSQL in a variety of ways. I might just build 2,
beefer systems that could replace the 4 web servers and handle the added
PostgreSQL overhead, and use a 2-way replication. That would also be
very straight-forward.
In reality, it's really the application. In fact, once you start
considering security, then it's all about the application.
E.g., I might not want one object to be on the same system that calls
another, "more privileged" object. I might want to use ticketing to
make systems re-authenticate on intervals. Especially if the systems
are on networks where there might be systems I don't control, or need
lower levels of access. Etc....
In reality, most web apps aren't so involved. Any MySQL approaches are
well-proliferated known. But PostgreSQL isn't that difficult to setup
and administer -- not even when you start talking 2-way replication.
-- Bryan J. Smith b.j.smith@ieee.org ------------------------------------------------------------------ Beware of advocates who justify their preference not in terms of what they like about their "choice," but what they did not like about another option. Such advocacy is more hurtful than helpful.----------------------------------------------------------------------- This list is provided as an unmoderated internet service by Networked Knowledge Systems (NKS). Views and opinions expressed in messages posted are those of the author and do not necessarily reflect the official policy or position of NKS or any of its employees.
This archive was generated by hypermail 2.1.3 : Fri Aug 01 2014 - 20:25:51 EDT