[pgpool-general: 6464] Re: Question regarding PG Pool high availability setup on Microsoft Azure

[Shereef Sakr]

Thanks, a lot for your response.

I see three options are listed:
1- Elastic IP, associating an IP with different machines according to which
machine is active. but has the drawback of being a public IP.
2- DNS, associating a DNS with different IPs according to which machine is
active,
3- Modifying a routing table on the network

the solution that I see can be feasible is using is #2 or #3 if I
understand correctly, but on the following conditions:

1- Do they work with internal IPs.
2- The changes are instantaneous, I'm afraid DNS might not fulfill this
option for example.

but I need to know what are the proper alternatives for #2 and #3 on Azure?
if someone has tested this before and knows if they fulfill the conditions
listed above that would help a lot.

And in general, doesn't anyone here have a recommended HA architecture on
Azure "apart from Azure services"? I suppose someone must have tried that
somewhere?

On Fri, Mar 15, 2019 at 4:46 AM Bo Peng <pengbo at sraoss.co.jp> wrote:

> Hi,
>
> On Fri, 15 Mar 2019 03:26:01 +0200
> Shereef Sakr <shereef.sakr at gmail.com> wrote:
>
> > Dears,
> >
> > Just a kind reminder, if someone can help me in this regard.
> >
> > On Wed, Mar 13, 2019 at 4:42 PM Shereef Sakr <shereef.sakr at gmail.com>
> wrote:
> >
> > > Dears
> > >
> > > I'm having a problem regarding a high availability setup on
> > > Microsoft Azure since PG Pool high availability depends on the
> existence of
> > > Virtual IP, which simply doesn't work on Microsoft Azure and is not
> visible
> > > except on the same virtual machine.
>
> I don't have so much experience with AWS.
> Below is a slide about setup Pgpool-II on AWS, but it is in Japanese.
> I hope you can find the solution to setting Virtual IP from page 38 to 46.
>
>   https://www.sraoss.co.jp/event_seminar/2017/jpug_20171103_pgpool_aws.pdf
>
> > > I found the example for Amazon AWS below, but I don't like the solution
> > > that it's a public IP visible to the internet, and we can't accept
> that for
> > > security reasons, also I didn't find a comparable example on Microsoft
> > > Azure too.
> > >
> > > https://www.pgpool.net/docs/latest/en/html/example-aws.html
> > >
> > > Also in addition to that I tried to disregard the virtual IP and
> configure
> > > my application using both internal IPs of both pg pool servers, but as
> per
> > > my observation I found the following:
> > >
> > > 1- Double the needed connections are opened to both backend application
> > > servers, which shouldn't be good to Postgres servers performance
> according
> > > to "PostgreSQL High Availability Cookbook" ( that Postgres performance
> is
> > > acceptable when the number of open connection is ~ triple the number of
> > > cores)
> > > 2- Also I find that both pg pool servers behave as Active/Active and
> can
> > > be used at the same time.
> > >
> > > So what is the recommended solution in this case?
> > >
> > > --
> > > Shereef Sakr
> > > ------------------------------------------------------------------
> > >
> >
> >
> > --
> > Shereef Sakr
> > Software Development Manager
> > Esri Northeast Africa
> > ------------------------------------------------------------------
> > Personal Blog: shereefsakr.blogspot.com
> > Technical Blog: fci-h.blogspot.com
>
>
> --
> Bo Peng <pengbo at sraoss.co.jp>
> SRA OSS, Inc. Japan
>
>

-- 
Shereef Sakr
Software Development Manager
Esri Northeast Africa
------------------------------------------------------------------
Personal Blog: shereefsakr.blogspot.com
Technical Blog: fci-h.blogspot.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.sraoss.jp/pipermail/pgpool-general/attachments/20190317/6431285f/attachment.html>