Thread: Multiple Server Installation/Design Question.
Hello,
I have been experimenting with the product, nicely done! I would like
to implement it in a live scenario now, however, I have questions.
I have 2 Data Centers, 1 with 2 ISPs and 1 with 1 ISP. I would like to
setup the product to replicate from 1 center to another - as close to
real-time as possible. I am sure I can find the steps to do this,
however, any information would be nice.
The question is, the Data Center with 2 ISPs, has a common Drive Array
[NFS based] and I would like to install the "engine" on each Blade
Server, however, have the same database files used by the engines. Is
this possible?
Details of Data Center #1:
ISP #1 - Windows Server, 1TB Drive - All Internal.
Details of Data Center #2:
ISP #2 - Windows Server, 4TB Drive Array - NFS Mount
ISP #3 - Linux Server, same 4TB Drive Array - NFS Mount
-- If Windows is an issue, I can easily put these on Linux (running
CentOS 5).
* My first task will be to implement a few tables with minimal
changes occurring, replicating between the two Centers, but again DC2
has 2 Internet Feeds. -- Think of the environment as a Wiki. Mostly
searches, with few changes or add.
* The next project after this one will be a Message Forum solution -
so it will have more writes then the Wiki solution - but still be a
search/view environment.
Unfortunately, I need to have a solution in testing this week - due
to the heavy snow storms up north, today is the first day I have had
access to the Internet to start asking question.
Thanks,
Ozz
On Mon, Feb 15, 2010 at 9:01 AM, Ozz Nixon <ozznixon@gmail.com> wrote: > Hello, > > I have been experimenting with the product, nicely done! I would like > to implement it in a live scenario now, however, I have questions. > > I have 2 Data Centers, 1 with 2 ISPs and 1 with 1 ISP. I would like to setup > the product to replicate from 1 center to another - as close to real-time as > possible. I am sure I can find the steps to do this, however, any > information would be nice. > > The question is, the Data Center with 2 ISPs, has a common Drive Array [NFS > based] and I would like to install the "engine" on each Blade Server, > however, have the same database files used by the engines. Is this possible? No, and often using NFS has its own set of issues to deal with, mostly to do with durability. > Details of Data Center #1: > ISP #1 - Windows Server, 1TB Drive - All Internal. > > Details of Data Center #2: > ISP #2 - Windows Server, 4TB Drive Array - NFS Mount > ISP #3 - Linux Server, same 4TB Drive Array - NFS Mount > > -- If Windows is an issue, I can easily put these on Linux (running CentOS Windows isn't the most performant OS for pgsql, but it works well enough for most situations. The real issue is that things like slony (replication) may not work as well or be as up to date, since it's developed on unix. > 5). > > * My first task will be to implement a few tables with minimal > changes occurring, replicating between the two Centers, but again DC2 has 2 > Internet Feeds. -- Think of the environment as a Wiki. Mostly searches, with > few changes or add. > > * The next project after this one will be a Message Forum solution - > so it will have more writes then the Wiki solution - but still be a > search/view environment. > > Unfortunately, I need to have a solution in testing this week - due > to the heavy snow storms up north, today is the first day I have had access > to the Internet to start asking question. The three replication systems most posted about and apparently in common use are Slony (complicated, versatile) londiste (simple, not quite as versatile) and bucardo which I'm not all that familiar with, but it looks interesting.
A while back someone answered a question I was running into - poor performance on count(*) from a table which is constantlygrowing. The answer was like looking at oracle's V$ table - and I could get a semi-current count. (I do not needthe exact count - just checking to make sure the table is growing). Ozz
Hello 2011/1/19 Ozz Nixon <ozznixon@gmail.com>: > A while back someone answered a question I was running into - poor performance on count(*) from a table which is constantlygrowing. The answer was like looking at oracle's V$ table - and I could get a semi-current count. (I do not needthe exact count - just checking to make sure the table is growing). http://www.commandprompt.com/community/pgdocs81/planner-stats-details column reltuples in table pg_class. Regards Pavel Stehule > > Ozz > -- > Sent via pgsql-general mailing list (pgsql-general@postgresql.org) > To make changes to your subscription: > http://www.postgresql.org/mailpref/pgsql-general >