FWH Setup and hardware issues

[Marc Venken]

After reading topics the last weeks in this forum and the active interaction with the members I decided to buy FWH.

I ordered today with Antonio and after some days, I will be up and running.

The program will be invoicing with a lot of Xbrowse use

For my hardware setup i have a question :

I have 5 Pc's in network (peer to peer) with 3 Pc's Win7 32Bit and 2 Pc's Win7 64 bit. (FWH32 I decided to buy, because I need 32 bit pc's for other software)

Now, What would give the best performence?

There is 1 lady that will be using the software almost all day long (90%) , the others will use it maybe 20/30 %

Do I put the data files on the pc of the lady and connect the others to the drive by drive letter ? sample -> P:\data
(she will have the highest speed, since the data in on her local drive)

Buy a Nas (https://maxict.nl/product/4776484/qnap- ... ts-431plus) and install the data on the nas, also with drive letters P:\data

Advantage for me, I'm sure the lady will forget to make dayly copies, but the Nas with 2 drives are clooned, so sure of backup.

What with lot's of statistical data

From My pc, I want to create and see lot's of statistical data. Seen samples her, It will be nessesary to connect to the data and create several indexes in order to use Xbrowse vieuws.

Will I create the cdx files on the lady PC or Nas, or localy on my system ? Speeds issues and waiting time to create indexes.

I'v even seen that cdx can be made in memory ? Will we ran out of memory or will it swapp to disk with large files?
I've seen rarely the use of cdx creation in memory, so I think there is a reason.

Regards,

[TimStone]

Good Evening,

You will get many viewpoints. I will share my experience.

My clients use multiple workstations. They run from 2 to 22 computers all sharing the same data. I suggest they need to use Advantage Database Server if they have more than 2 or 3 computers on the network, but some do not.

Their "server" is either a dedicated computer that does nothing but share the data, or an office computer that is used very lightly but it always running. If the lady is doing most of the work, and has a fast computer, it will work to have the "server" files on her computer.

Indexes are important, and you can have many to speed your operations. I use the SCOPE commands a lot rather than filters, and they do require an index to work properly. I maintain multiple indexes on each file, but there is no real time delay in doing this, and using them does speed up your operation a lot.

My clients run all of their computers all day long, and are constantly interacting with the data. Peer to peer has it's issues, especially if the computers are Home OS rather than PRO. The networking is just much weaker. ADS removes that problem.

As you work with this, feel free to ask questions here. The answers will come quickly.

[Marc Venken]

Tim,

ADS is a software solution for better performance as far as I see. Is there also new hardware needed, or better Desktop systems ?

Will the way of programming in FWH be different ? Have you a small snipped of code how a dbf is opened and a scope is set for a Xbrowse.

With ADS it's the way of SQL systax like in PHP? then i've seen samples in the Forum, and will I look into it.

[TimStone]

Marc,

The Advantage Database Server has two modes of operation. It has "local" and "remote". Local is free and is basically support for peer to peer. Although that is native to the DBFCDX RDD, if you use the LOCAL drivers, then all you have to do is add the actual Client Server software and change the method to REMOTE and you will have true CS operations. This means that data is all processed on the "server" rather than passing larger packs of data, and indexes, to each client machine. It is also significantly faster.

ADS could be SQL, but you can use it for normal DBF operations, which is what most people do. Thus, if you are experienced with Clipper programming, ( xBase ), you will essentially add a few lines of code to select the ADS RDD, and otherwise everything is pretty much the same.

I would suggest you consider, from the start, the use of data objects. You will find samples of tDatabase in FWH, and there is an available extension to the class, tData ( plus tRecord ) from James Bott that adds even more power. Instead of needing work areas, objects handle all of that for you. The syntax is also cleaner. Data is stored in buffers first, and you change their content, then use a save( ) to commit the changes. No more long REPLACE lines. Locking at the record level ( or file ) is handled automatically as part of the class. This will reduce a lot of code, and confusion. You will find examples in the samples folder of FWH.

Antonio probably suggested you use Harbour. Some people use Borland, others use xHarbour, and some use Harbour. I actually have all my code written so I can do builds with xHarbour ( the .com version includes the compiler and linker, but does cost extra ), and with Harbour ( using MSVC++ 2015 which is free with Visual Studio 2015 Community edition ). Once you decide on your "build tools", you will progress quickly.

Of course, you do need to understand xBase programming. You have not mentioned your background, but I assume you have existing code to convert. This forum will have many topics on skills you will want to learn. Just do a search.

Tim

[Marc Venken]

I will look into the free Local ADS. I suppose there are several companies for this kind of software. Witch do you use ?

Normal dbf with cdx will be my choise, since I see the old code I have (16 bit clipper) with many TSbrowse6 (some time ago). This kind of code is readable for me and with the samples i've seen in the forum I can start a new software. (The old is not good coded if I see the samples on the forum.

I suspect the software from Antonio in the next days, and then I will also have access to the samples you mention.

The data objects are good indeed. I have read the doc part I from James and i'm convinced to use it. Lot's of replace... in the old source.

Marc

[Adolfo]

Marc.

Is this a brand new development ?
If yes... why don't you use SQL, (MYSQL or better MariaDb) much faster, reliable than ADS and also FREEEEEEEEEEE (MariaDb). There are TONS of help in the web, you can use Tmysql, Ado, TDolphin

My 2 cents.


From Chile
Adolfo

[TimStone]

The ADS LOCAL software is available and will not be a problem. However, it is very simple to add later. For now just use the DBFCDX driver.

There was a suggestion to use SQL. I thought about it, but stayed with DBF for the following reasons:

1) BLOAT - I converted my data files from dbf to SQL and the new database was six times larger than my original.
2) RESTORE - I have an auto backup routine built in that makes a zip archive of all .dbf/.fpt files each night. I have 128 .dbf files. Some of them are small and have small amounts of data. Others are large. If one file gets damaged, I can restore just that database and perhaps the amount of data that must be re-entered could be from 0 entries to a few. With SQL I would have to restore the whole database, and that could mean a whole lot of data re-entry.
3) UPDATE - I have routines to enhance a single data file ( ie. add fields, change field lengths ) without using any others. This may not be as easy to do in SQL
4) REPAIR - I have tools that can actually fix .dbf files. Sometimes the repair is very simple ( false EOF ), or it can include removing garbage data. I have not found "repair" tools for SQL. The response I get is "restore it from the last backup".

Finally, maybe your "coding" was not the best, but the design and layout of the program may be good, and the client is used to that format. You might be able to work from your original source and improve on it. When I converted from DOS to Windows, I used /* */ to comment the original code, pushing it to the bottom of the .prg file, and then added the new code above it. Then I could copy / paste sections from the old to the new area, and modify them as needed. It worked out well.