PostgreSQL & VMWare

Hi,

I work on a project that collects geolocalized data.
All data will be in PostgreSQL / PostGIS.

The small PostgreSQL databases will be on Linux guests on VMWare hosts.

The size of the main database will grow by 50 TB / year, 500 M row / day.
For the largest one, we plan to test different options.
One of them is to stay with Linux on WMWare.
Outside the questions about schema, sharding, I would appreciate if some of you have informations, benchmarks, stories about big PostgreSQL databases on Linux guests on VMWare hosts.

Regards,
Jean-Gérard Pailloncy

--
Sent via pgsql-general mailing list (pgsql-general@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-general

On Fri, 3 Jul 2015 12:35:07 +0200
Jean-Gérard Pailloncy <jg.pailloncy@adnow.fr> wrote:

I work on a project that collects geolocalized data.
All data will be in PostgreSQL / PostGIS.

The small PostgreSQL databases will be on Linux guests on VMWare hosts.

The size of the main database will grow by 50 TB / year, 500 M row / day.
For the largest one, we plan to test different options.
One of them is to stay with Linux on WMWare.
Outside the questions about schema, sharding, I would appreciate if some of you have informations, benchmarks, stories about big PostgreSQL databases on Linux guests on VMWare hosts.

The place I'm working now did a feasibility study about installing
their primary app on vmware instead of directly onto the hardware.
Their conclusion was that the app would be about 25% slower running
on VMWare. The app is very database-centric. However, I wasn't
involved in the tests, can't vouche for the quality of the testing,
and there _are_ other pieces involved than the database.

That being said, I've used PostgreSQL on VMs quite a bit. It does
seem slower, but I've never actually benchmarked it. And it's never
seemed slower enough for me to complain much.

The concern I have about running a large database on a VM (especially
since you're asking about performance) is not he VM itself, but all
the baggage that inevitably comes with it ... oversubscribed hosts,
terrible, cheap SANs, poor administration leading to bad configuration,
and yet another layer of obscurity preventing you from figuring out
why things are slow. In my experience, you _will_ get all of these,
because once you're on a VM, the admins will be pressured to host
more and more VMs on the existing hardware and/or add capacity at
minimal cost.

There's nothing like a VM where you never know what the performance
will be because you never know when some other VMs (completely unrelated
to you and/or your work) will saturate the IO with some ridiculous
grep recursive command or something.

--
Bill Moran

--
Sent via pgsql-general mailing list (pgsql-general@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-general

På fredag 03. juli 2015 kl. 12:35:07, skrev Jean-Gérard Pailloncy <
jg.pailloncy@adnow.fr <mailto:jg.pailloncy@adnow.fr>>:
Hi,

I work on a project that collects geolocalized data.
All data will be in PostgreSQL / PostGIS.

The small PostgreSQL databases will be on Linux guests on VMWare hosts.

The size of the main database will grow by 50 TB / year, 500 M row / day.
For the largest one, we plan to test different options.
One of them is to stay with Linux on WMWare.
Outside the questions about schema, sharding, I would appreciate if some of
you have informations, benchmarks, stories about big PostgreSQL databases on
Linux guests on VMWare hosts.
 
We have a 1.5T and growing DB which we started out hosting on VMWare, BIG
mistake!
 
Never, ever, run a production DB on anything other then bare metal servers.
Once we blasted VMWare's 8GB disk-cache our DB random-read speed went down to
1MB/s with 8K blocks (as PG uses), yes that's one megabyte per second!
 
-- Andreas Joseph Krogh
CTO / Partner - Visena AS
Mobile: +47 909 56 963
andreas@visena.com <mailto:andreas@visena.com>
www.visena.com <https://www.visena.com&gt;
<https://www.visena.com&gt;

Hi, It is slower in default settings of vmware environment. But today
vmware has invented other things like SSD cache, vsan etc. If you can
afford that kind of i/o improvments in your environment, it is very
reasonable to continue on vmware, especially due to the ha things.

2015-07-03 14:37 GMT+03:00 Andreas Joseph Krogh <andreas@visena.com>: