Advise requested and suggestions welcome
От | Franz J Fortuny |
---|---|
Тема | Advise requested and suggestions welcome |
Дата | |
Msg-id | 6f4C5.2166$Ow4.108144@news-west.usenetserver.com обсуждение исходный текст |
Список | pgsql-general |
This is one of those “help me decide” petitions. Our company requires a database engine capable of handling at least 200 connections simultaneously. The tables in that database are interrelated to the extent that it would not be practical to divide them in more than one physical database. The size of this database will increase on a daily basis. We would like to keep, on line, at least 2 years of our operations. Our calculations indicate about 9-10 GB of disk space will be required. We will want to keep in “read-only” databases old information, ready to be used for decision making applications. We will most likely be using C++ Builder 5(+) for our applications. We would like to be able to count on VCL objects that have been exhaustively tested in actual applications. We have not had a good experience with a commercial database (SQL) server. The product works fine most of the time, but it does fail unexpectedly once in a while and we have not been able to get the company to react in the direction of fixing the bugs causing the sporadic failures. There are at least two “open source” SQL servers that we have been testing: Interbase and PostgreSQL. The C++ Builder 5 includes the IBX objects, and using these objects our applications would not even require the installation of packages like BDE or even ODBC drivers to connect to our Interbase servers. (It is our intention to host our Interbase or PostgreSQL servers in Linux machines, with several CPU’s and database serving dedicated). Though C++ Builder 5 does not include VCL objects for PostgreSQL, we have found at least one free set of VCL’s (Zeos DBO) that would allow our applications to install in client machines without BDE or ODBC. The big question is: are either Interbase or PostgreSQL SAFE? Our commercial product has been SAFE, in the sense that after two years, we have lost only 25 minutes of operations, due to corruption (generated by the program itself) on its own recovery log files. Why would we want to abandon such a safe SQL Server for another one whose level of safety we don’t know? This is where you, the community of PostgreSQL and Interbase programmers and users, can help us. Have we had a good, safe performance compared to what you know from your daily activities? Do you know of successful operations in PostgreSQL or Interbase where operations similar in dimension as ours are being run? Would you say that 200 simultaneous users, several thousand (5-10) committed transactions daily and 25 minutes lost in 2 years is a good, regular, expectable, better than average or less than average considering that the database server DOES FALL dumping core files 3 or 4 times a month? Would you stick with this product or would you try to find one that will not fall generating core files and die 3 or 4 times a month? Would you say it is “impossible” to get zero down time in this kind of products (SQL servers)? Your comments will be highly appreciated. F J Fortuny
В списке pgsql-general по дате отправления: