Re: PostgreSQL still for Linux only?

Tope Akinniyi wrote:

> Hi all,
>
> In my country Nigeria (and even African continent), we do not eat what
> the western world eat. We wear different styles of cloths. In the same
> vein, our computerisation culture is different.

Having lived in Indonesia, I can sympathize with your situation. It is
not just Africa, but most of the developing world.

>
> I must submit that computers became popular in Nigeria by Windows
> desktop system. While the western world were exposed to *NIX from the
> beginning, we were introduced to computing via DOS and later Windows.
> That is our IT antecedent and culture. People use database engines
> such as Oracle, Firebird, Sybase, mySQL, etc on Windows here and they
> manage them and survive. If because you want to recommend PostgreSQL,
> you insist on Non-Windows OS, the first question clients ask you is
> why is your own different? Why must I switch from Windows to *NIX
> because of your PostgreSQL? You might end up not succeeding in that
> bid. And we are used to the blue screen (crashes) and each IT house in
> Nigeria has gone the extra mile to ensure the safety of the operations
> of its clients. Everyone is a product of his environment,
> peculiarities and experiences.

If you want a reasonable open source RDBMS for production use on
Wondows, I would suggest that you use Firebird. However if Windows is
not the selling point, consider the following:

1) You may be able to get extra use out of older systems by installing
Linux and PostgreSQL. This may perform better than Windows and Firebird
as long as you don't need a GUI. This may be more reliable than Windows
especially if you can't afford high-end hardware (ECC RAM, SCSI drives,
etc) for your production servers anyway.

2) The PL's available for PostgreSQL add a lot of flexability.

>
> As an IT organisation that wants to stay in business you need to give
> to people what they wants. I think that is the basis of service. I
> have some deployments of PostgreSQL on Windows servers. I must admit
> that we have not had any problems so far.
>

The glory of open source is that people will do what they want with
it. PostgreSQL for Windows is not really something I would run a large
production database on at the moment. However, open source tools tend
to develop in strange ways. I am sure that as PostgreSQL on Windows
becomes more popular, the issues will get worked out as much as possible.

> Notwithstanding, due efforts must be made to protect your clients'
> operations whether you use Windows or Posix. In that regards, I
> thought of reducing the risk factor by implementing replication on
> some of the servers.

Command Prompt's solution works on Windows. Slony will require some
porting, but if this is important, you can hire a programmer to help
with the porting :-) Otherwise you can wait for someone else to do it.

>
> I sought Windows replication tool for and could not get. I checked
> PgFoundry and the one there put a banner and said NOT FOR WINDOWS.
> Then I said is this PostgreSQL for Windows a joke? That prompted
> my post - IS POSTGRESQL FOR LINUX ONLY?
>

Check the archives about Slony-I and Windows. Maybe ask the developers
how much work it would be to port it. If labor is inexpensive in
Nigeria, maybe you can hire a programmer to do it.

> Now, as the CEO of an IT organisation, I want to draft my final
> blueprint on PostgreSQL. I need your advice on this.
>
> 1. If I can manage it, can I continue to use PostgreSQL on Windows and
> watch as it evolves? I recognise the points certain respondents made
> on earlier; which was PostgreSQL on Windows is still a baby boy, do
> not expect it to walk like a man or expect it to possess the features
> of a man.

Ok, maybe others can provide more refined estimates, but....

I expect that it will be 1-2 years before PostgreSQL on Windows is
mature enough for higher-load purposes. You can however help by using
it, and communicating your experiences with programmers. If this is not
enough, you can even pay someone to fix things for you. These are
selling points of open source software.

>
> 2. This response is alarming:
> Tom Lane wrote in digest V1.5092:
> >We are supporting Windows as a Postgres platform for the benefit of
> developers who want to
> >do testing on their laptops (and for reasons best known to themselves
> feel a need to run >Windows on their laptops).
>
> a. Who are the 'we' Tom is talking about?
> b. Is he speaking for PostgreSQL Developers and the entire PostgreSQL
> community?

As much as I don't like to speak for others, I read this as saying
something like:

"We (the core developers) began work on the Windows port because we
wanted to support developers running PostgreSQL on their systems."

> c. Does this mean that PostgreSQL for Windows is just a toy or model -
> Oh do not take it serious? Or is the Windows version by design a
> miniature of the *NIX version, lacking the requisite mechanism of a
> reliable database?

I think the core team takes all aspects of PostgreSQL very seriously.
Part of the problem is that they are so serious about it that they don't
want problems in Windows to smear the name of PostgreSQL. So comments
like Windows being unsuitable for any RDBMS use applies to MS SQL and
Firebird as PostgreSQL. But that doesn't mean that people won't do the
best that they can to make it work.

> d. And does that mean the developers can decide to withdraw
> development and support for the Windows version anytime they so wish?
>

Individual developers can decide whatever they want. However, as long
as Windows/PostgreSQL is popular, it will be supported regardless of
whatever the core team wants to do. This is because the community has
access to the source code and that the community will include a large
number of software developers. But no, I don't see the core team
deciding to remove supported platforms any time soon. We are not a
centralized commercial enterprise like MySQL, so whatever the community
really wants, they will eventually get.

> I am not against Linux or any Posix for any reason. In fact one of my
> two office servers run Mandrake Linux. But I am grateful that
> PostgreSQL recognises the fact that we all can and will not be in the
> same boat. So it is good to support many boats.
>
> Tom lane's post is worrisome to me. It bothers on consistency. Would
> PostgreSQL be consistent for Windows? If not, I think at this stage I
> can easily roll back and migrate my clients back to other Windows
> Database system where I feel I will be secured for some time to come
> as using PostgreSQL does not affect much of my operations. I am just
> expanding my varieties.

Why should it matter? What will happen is that PostgreSQL on Windows
will become commonly used in both development and production
environments. It won't be fast like on Linux because of fork() overhead
but it will be supported as long as the community wants it.

>
> I think managing PostgreSQL on OS I desire should be my own duty. The
> point is that PostgreSQL can be available for what I choose to use it
> for and where I choose to use it. Managing failure points of my OS
> should be left to my technical expertise. Well if I can get some
> support from some sources, fine.
>
>
> I do not seem to be comfortable with this "Windows will spoil
> PostgreSQL reputation position" as posted by Schroeder. Is PostgreSQL
> the only database engine running on Windows? There are million of
> licences of Oracle, mySQL, Sybase, etc for Windows servers. The
> company that uses them are up and running; not as if only
> organisations running DB on Posix are existing. Who blames mySQL or
> Oracle when it crashes on Windows OS? If PostgreSQL cannot thrive
> where others thrive, it will be quite unfortunate. You cannot shut
> yourself indoors because you anticipate a rainfall (that might not
> come). What would be the empirical basis for our judgement if
> PostgreSQL is not used on Windows? Crashing MS Office on Windows is a
> different situation from what you would get running PostgreSQL. I do
> often witness many utility *NIX applications do crash on our Mandrake
> server, but not PostgreSQL crashing.

As I said, the core team takes the security of your data very
seriously. Note that this does not mean that it will nto be supported,
but just that people don't think you should consider doing this.

>

I hope this response is helpful.

Best Wishes,
Chris Travers
Metatron Technology Consulting