December 23rd, 2002, 02:00 PM
Help With PGSQL Implementation Proposal
I have just begun drafting a proposal to replace Microsoft Access with PostgreSql as our database system for serving Intranet content to our in-house users here at work. I have been poring over the PostgreSql documentation and Advocacy sites for information to help further my cause. I would also like to poll the users of DevShed for any insight they may have into this project. I would like to hear any suggestions on what I should focus on or include in my proposal to get this done. Here's a little background on what I'm working with / against:
We are developing for a Customer Service Intranet. At any given time, there are approximately 120 people on it at one time. There are not, however, necessarily 120 simulataneous database connections. The platform currently is a WinNT 4 system with a possible future upgrade to XP. We are serving over 10/100 LAN, using perl for server side development and using Access in our production environment.. this last point is the problem and scares me.
This is a Microsoft Shop all the way. If it's not Microsoft, it's tough to convince them to go with it at times no matter the logic.
I did, however, manage to push perl onto the production server instead of ASP.
Our department (and, the whole company as a global entity for that matter) has recently been fixated on cost-cutting like so many others lately.
We already have Sql Server and Crystal Reports. I do NOT know how many licenses we have or would need to purchase if we moved onto Sql Server instead.
I will first be trying to convince my boss who does not have a great deal of expertise of server-side technologies (he's a graphics guy and he seems quite happy to stay that way ). I will then be dealing with administrators who may or may not have any technical expertise on the subject of databases, intranets, and/or server-side programming.
We have a budget - we just choose not to use it, and I don't know what it is...
All that said, I figure I will push the initial implementation costs as a one time thing that beats the overall licensing costs of Sql Server and Oracle (I don't think I can get access to the Oracle server though..). I'll try and get some estimates to compare the overall TCO of PG vs. Sql Server. Of course, I'll also focus on reliability and scalability as well as the overall performance of PG vs. Access (no contest..). I'm going to point to some recent trouble in which we had some databases corrupt in Access and the excessive difficulty that I experienced trying to locate correct and complete documentation for programming Access. I'll also mention the kludges (without saying kludges) that were implemented as a result of limitations in the Access database and the instability that resulted from them. The focus, however, will be on PostgreSql's strong points, no Accesses numerous shortcomings.
Any further input would be greatly appreciated as I'd really like to push a better database into my job so that I don't have to keep fighting with Access....
December 23rd, 2002, 05:49 PM
The only thing I can do is wish you good luck - you'll need a lot of it.
See, thing with MS products is that they perform just great on all-microsoft stack, which you said your company is or "wants" to be.
I could search for some info if you clear this up - are we comparing access to postgres or sql server to postgres? From what I see your company sees that access is no good, so probably proving that postgres is better then alternatives would be more efficient then proving that it is better then what you have right now.
And you know I mean that.
December 23rd, 2002, 07:19 PM
Thing is this... we've been given a project that is the "pet" of the VP of the department. Trouble is.. he's about as involved in it as I am in the war in Afghanistan... so we get to throw our weight around on threat of going to the VP for clearance and angering him when it comes to pushing around troublesome people in OUR department, but, since we can't REALLY go to the VP (we can just threaten..) we can't push around anyone outside of our department... that includes the IT folks who control all of the technologies we use. Therefore, we usually just wind up defaulting to whatever technology is most available.. in this case, that was Access for the database. Back in the day that I first suggesting serving some of our content from a database I didn't have the clout built up to argue against using Access. But, now I've implemented several project successfully, so I'm going to give it another shot.
I don't know or even care if we can get access to Sql Server or Crystal Reports, however. So, basically, what I'm going to try and push is "Why we should install pgsql and dump Access in the farthest corner of the galaxy - BUT - only for our department.. everyone else can still use Access if they feel some compelling urge to corrupt all their data".
But you make a good point - if I tear Access to pieces they'll just say "What a coincidence - we have a Sql Server install that you can use....". BUT... I don't know. I suspect they probably purchased Sql Server as part of the old licensing scheme... so if they added us, they'd have to use up and/or purchase more seats for it (as opposed to the M$ web licensing where they license to a CPU rather than the number of users). So...what I need to do is prove that:
1. Access is too junky for our application.
2. Sql Server is overkill for it.
3. Who the hell uses Crystal Reports anyway?
I'm also considering mysql btwl... I myself haven't really been able to determine whether one is really better than the other to be honest...
And speaking of Sql Server performance... this was interesting.. though, it was a "benchmark", so I'm offering free very-large-grains-of-salt with this link:
Last edited by Ctb; December 23rd, 2002 at 07:23 PM.
December 23rd, 2002, 07:22 PM
If you want to use postgresql on a windows system then I would suggest rethinking your solution. The problem is that the current postgresql code uses the cygwin environment under windows and isn't something you want to use in production. The next release will apparently run in a standard windows environment and be a lot better under windows.
If using a unix system for postgresql is acceptable then everything should be okay. A good way to get support for using postgresql is to compare the licensing costs of using sql server or oracle versus postgresql. You can use the fact that postgresql is used to hold the information for the .info and .org dns domains as well as pointing to some of the other areas where postgresql is used to support your contentions that it can handle the load. If support will be a problem there are several companies that provide support contracts for a postgresql server including Red Hat.
December 23rd, 2002, 09:05 PM
fplscheme - agreed. The necessity of cygwin is one thing that may prevent me from ultimately choosing to push pgsql over mysql. However, at the moment I'm holding both of them as potential candidates because I think that once pgsql is available natively on Windoze, it will be a viable solution.
I've checked out licensing and support costs. There's simply no contest. $5000 per seat for Sql Server or $200 per processor for MySql.. no contest at ALL.