Re: [Dataperf] Flyweight Database

Thu, 14 Sep 2006 14:31:56 -0700 

Hi Bruce,
I agree it would be a tough road, I work in both spheres, and I love working with DP because it is so robust and the development time in the database is so quick... Even when I am suggesting DP as a backend web application, where the DOS interface is not even remotely exposed to the user, the people who want to talk technically with you are a very hard audience. 


Before the Java port, I think in the web database sphere, DP would be more 
readily accepted if it could run thru IIS on Windows. I have only ever been 
able to get it to run on Apache under Windows...



I have been putting my DP web apps on a Linux server running DOSEMU and the 
best I can manage for loading DP, (doing nothing) and then exiting is just a 
tad under 3 seconds and that can seem like a lifetime (and then you have to 
add the DP report code), many database systems deliver web performance akin 
to just standard file serving from a web server.   (by the way Bruce, how do 
you get such great pewrformance from your AJAX (AJAT) zip code sample 
application?  Are you using Firestorm code to drill directly to the data)



I love the DP keep a total, and the collateral things you can achieve with 
it (ie automatic creation of necessary related records etc)... but two phase 
commit transactions go a lot further...  For example in a double entry 
accounting system, you will want to post a minimum two separate records to 
make up entry...  but it is absolutely imperative that if something doesn't 
work, that neither record is posted. But it is not just financial system, 
the same need for multiple transactions to be posted as a seemingly atomic 
unit occurs all over the place.



SQL has its downside and its blessings...  On some of my larger systems 
there are many hundreds of SQL Views/Queries and having to write a separate 
DP report for each of those would be daunting. Both SQL and DP have things 
the other can't do...  Try creating an unknown number of columns pivot table 
like report in DP, or even a sorted aggregation, which in SQL is a piece of 
cake...



I would love for DP to be reinvigorated, it has so much to offer, but as you 
said, the world has probably already voted on database technology.


Regards
Brian


----- Original Message ----- 
From: "Bruce Conrad" <[EMAIL PROTECTED]>

To: "DataPerfect Users Discussion Group" <[email protected]>
Sent: Thursday, September 14, 2006 4:13 PM
Subject: Re: [Dataperf] Flyweight Database



Hi Heman, Brian, and others,


I agree with your assessment of the value of DP technology. The main 
difficulty marketing DP to the industry is that the industry has already 
"made up its mind" on some issues, and is not willing to accept innovation 
that is incompatible with those decisions. Three major decisions come to 
mind immediately: that it must use two phase commit, that it must run as a 
separate service, and that it must use SQL for communication between the 
database server and applications.



Aside: The reluctance to accept innovation is very curious, because in 
most other software areas, newness and innovation are viewed as essential 
features. We joke about being on the "bleeding" edge, and one site even 
calls the open source software it hosts "fresh meat". In these other 
areas, your code HAS to be new to even be considered. Yet, in the database 
world, new techniques are abhorred, and will not be seriously considered.



Over the years, as I have contemplated this, I have developed answers to 
each of the firm decisions of the industry.



Two phase commit? Keep-a-total handles most needs very cleanly. For 
others, a change in philosophy removes the need for it. For example, a 
system might use a transaction to create an invoice and all of its line 
items. In DP, we would just create an empty invoice and then add its line 
items to it one at a time, with KAT for a running total. No transaction is 
needed.



Separate service? The amount of code required for DP is so small, this is 
not a big issue, and with DP's imperceptible use of network bandwidth, 
multiple clients can operate on the data concurrently, with high 
performance. Of course, applications DO have to be written to deal with 
the possibility that another user might have changed/deleted a record. 
Experience with DP shows that this happens rarely and that users just 
accept it and deal with it.



SQL? With SQL, the application gets multiple rows (records) from the 
database server, and then has to turn around and iterate through them. How 
much cleaner it is to position to the first record in (a subset of) the 
table (i.e. panel) and then process the records one at a time. Application 
programming with SQL is very tricky. Using the DP engine, it is extremely 
simple.



Brian also brings up the problem that the code does not run on the most 
popular operating systems for web servers. This is a very good point, and 
my response would be two-fold. To first of all port it to a language such 
as Java which is available everywhere. Secondly, IF NEEDED, (parts of) it 
could be translated from Forth into Java or other languages for 
performance reasons. Once this is done, it will run on all web server 
systems.



Once in Java, the "multiple loading" problem just goes away, as DP would 
become part of the servlets defining the application, and these are loaded 
only once and then used in any number of requests.



As is often the case, I believe the technical problems are far easier to 
address than the social problems. The new DP will have to be marketed as 
if it were something totally new and innovative. Even so, acceptance by 
the marketplace seems problematic.



Why not just add SQL to the DP engine, let it run in a separate service, 
and give it two phase commit? Because this is a WHOLE lot of work--at 
least one order of magnitude more work than porting to Java. Besides it is 
not compatible with many of the things that make DP great--they simply 
cannot be expressed in the SQL language.



I welcome this opportunity, and would be interested in being a part of 
anything that develops.


Best wishes,
Bruce

Heman wrote:

Thanks Brian.  I’ve not commented, hoping on some more active interaction 
from others in the group.  I see the value of the DataPerfect “engine” in 
the same vein as MySQL.  A robust, “flyweight” engine that can be used as 
a viable, effective alternative to the big players.  The pluses of 
DataPerfect we are very familiar with, and they include its small size, 
its data integrity model, and its speed.  I know that some real work will 
exist to finally deliver a world-class competitor in a real-world 
business, but the very reason this discussion is happening is to get 
others feelings about the value, and excitement about such an idea.  The 
very definition of what that final product would look like as a 
“flyweight database” is not settled, but I believe it will be exciting to 
find out.



 I have always believed that the core ability and technology of the DP 
engine are a hidden jewel, deserving more than a quiet rest in piece in 
obscurity.



 Here is an update.  I just got off the phone with the VC firm interested 
in this concept, and they have asked me to submit a formal proposal.  I 
will work with Lew on that.  But I welcome all comments.


 Heman

 ------------------------------------------------------------------------


*From:* [EMAIL PROTECTED] 
[mailto:[EMAIL PROTECTED] *On Behalf Of *Brian Hancock

*Sent:* Wednesday, September 06, 2006 1:12 AM
*To:* DataPerfect Users Discussion Group
*Subject:* Re: [Dataperf] Flyweight Database

 Hi Tony and others,


 In regards to the flyweight database, Hernan asks is there any place for 
a non-Windows database.



 Modern database products are not driven in Windows they are driven thru 
the web, and as such they basically transfer text back and forth, and the 
concept of Windows or non-Windows is fairly academic. MySQL got to the 
popularity it has, despite initially not having a graphical user 
interface to manage it, later on it got an admin console, but initially 
it was bare...  When you use MySQL the interface is what you make of the 
client front end....  DataPerfect can already be considered a database 
without an interface when you use it across the web, it then has a bonus 
of being able to use it in text mode, to define applications, thats far 
far better than MySQL or other similar databases.



 I see two problems with DataPerfect in the web space...  Firstly it does 
not natively run on any of the main operating systems for web servers or 
database servers, and two that it runs as a CGI rather than as a service, 
which means there is costly time to load it up each call to it, and 
simultaneous calls, requires multiple loading of all the code into 
memory.  Having said that, DPWeb is great, as low stress applications can 
be built very quickly and robustly. In terms of financial applications, 
where you have many components forming a transaction, the lack of two 
phase commit, (either they all get saved or none gets saved) would be a 
worrying in terms of making a web based financial system.


 Regards

Brian

 ----- Original Message -----

    *From:* Tony Perez <mailto:[EMAIL PROTECTED]>

    *To:* 'DataPerfect Users Discussion Group'
    <mailto:[email protected]>

    *Sent:* Wednesday, September 06, 2006 4:11 PM

    *Subject:* RE: [Dataperf] Flyweight Database

     Heman & Lew ... this is a very interesting discussion that I hope
    many DP'ers will chime into. Draper is a very well known and
    respected name in the VC community. If they are calling out for a
    "flyweight database", there must be a reason. A saying in Spanish
    states "cuando el rio suena, agua lleva" ... "when the river sounds,
    it carries water".

     Coincidentally with this "nugget" are the article in the August 16,
    2006 issue of Forbes regarding Larry Ellison's stronghold on Oracle
    and an assignment I've been working on since May.

     I encourage all DP'ers to read the Forbes article as it addresses

    issues that open windows for applications such as we all grasp as 
DPWeb.


     As to the assignment, I consult with a developer of manufactured
    home parks. These parks are gaining popularity as affordable housing
    in areas such as California where home prices are out-of-reach for
    the great majority. I was asked today to put together a "billing
    system" using QuickBooks. I can plough through QB and get somewhere
    with the billing system. But I know, I can do more for our
    long-term needs and growth using an internet version of DP.

     I cannot forget those wonderful uses of DP we saw at the Conference
    in Huntington Beach ... coincidentally ... Labor Day weekend 2004
    ... and the enthusiasm the collateral interfaces of the tow truck
    operation in Arizona and Colin's and Brian's brilliant work in the
    southern hemisphere produced.

     I am definitely interested in thrusting my development efforts to a
    DPWeb solution. I look forward to our community's comments.

     All the best to all,

    Tony

     -----Original Message-----
        *From:* [EMAIL PROTECTED]
        [mailto:[EMAIL PROTECTED] Behalf Of *Heman
        *Sent:* Tuesday, September 05, 2006 4:35 PM
        *To:* 'DataPerfect Users Discussion Group'
        *Subject:* RE: [Dataperf] Flyweight Database

        My guess is, and I’ve not visited with these guys directly yet,
        though I’ve sought audience, is the handheld market, embedded
        markets, etc.

         Heman

         
------------------------------------------------------------------------

        *From:* [EMAIL PROTECTED]
        [mailto:[EMAIL PROTECTED] *On Behalf Of *Don
        Friedman
        *Sent:* Tuesday, September 05, 2006 5:26 PM
        *To:* DataPerfect Users Discussion Group
        *Subject:* Re: [Dataperf] Flyweight Database

         Heman,

        Great to see you around. Been a long time.

        I wondered whatever happened to the flyweight database market.
        Is anyone out there making money at it? And what makes us think
        that a non-windows DP could have a place there?

        Don Friedman
        Pittsburgh, PA

        On 9/5/06, *Heman* <[EMAIL PROTECTED]
        <mailto:[EMAIL PROTECTED]>> wrote:

        Guys, I just had a fun conversation with Lew.  He and I
        discussed an article in the September issue of Business 2.0, the
        20 Smartest Companies To Start Now.  Number 2 in the article is
        a VC looking for a flyweight database.  The article isn't posted
        online yet as I write this, but could be shortly.  Lew and I
        would like any and all comments back regarding DataPerfect's fit
        in addressing the need stated here.  Bruce & Thom, would you two
        comment and engage on this discussion in particular?  Thanks,

         In a nutshell, the article states: Draper (the VC) loves
        startups that can upend corporate giants with simple products
        and cheap technology.  Oracle, IBM, and Microsoft have had a
        stranglehold on the $13.8 billion database business for more
        than a decade, and while newer players like MySQL are making a
        dent, Draper thinks there's an opening for a startup that can
        deliver most of the benefits of standard Big Blue products
        without millions of lines of code or an army of consultants and
        IT managers.

         This little article blurb lends me to think (and Lew as well)
        that DP would play well here.  We intend to discover the reality

        of that. So please chime in and give some input and feedback. 
I've started my investigation to discover more specifically what
        this guy is looking for, and preparing for a presentation to 
them. Thanks,


         Heman

        +1.801.368.9398

         --
        No virus found in this outgoing message.
        Checked by AVG Free Edition.
        Version: 7.1.405 / Virus Database: 268.11.7/437 - Release Date:
        9/4/2006


        _______________________________________________
        Dataperf mailing list
        [email protected] <mailto:[email protected]>
        http://lists.dataperfect.nl/mailman/listinfo/dataperf





        -- 
        Don Friedman

        ProfessionalRecords.Com LLC
        PRS Data Systems
        205 S Main Street
        Pittsburgh, PA   15215
        412-784-1600 - 1-800-PRS-FILE
        412-784-1615 Fax

        --
        No virus found in this incoming message.
        Checked by AVG Free Edition.
        Version: 7.1.405 / Virus Database: 268.11.7/437 - Release Date:
        9/4/2006

         --
        No virus found in this outgoing message.
        Checked by AVG Free Edition.
        Version: 7.1.405 / Virus Database: 268.11.7/437 - Release Date:
        9/4/2006

    ------------------------------------------------------------------------

    _______________________________________________
    Dataperf mailing list
    [email protected]
    http://lists.dataperfect.nl/mailman/listinfo/dataperf


--
No virus found in this incoming message.
Checked by AVG Free Edition.
Version: 7.1.405 / Virus Database: 268.11.7/437 - Release Date: 9/4/2006


--
No virus found in this outgoing message.
Checked by AVG Free Edition.
Version: 7.1.405 / Virus Database: 268.12.3/446 - Release Date: 9/12/2006


------------------------------------------------------------------------

_______________________________________________
Dataperf mailing list
[email protected]
http://lists.dataperfect.nl/mailman/listinfo/dataperf

_______________________________________________
Dataperf mailing list
[email protected]
http://lists.dataperfect.nl/mailman/listinfo/dataperf



_______________________________________________
Dataperf mailing list
[email protected]
http://lists.dataperfect.nl/mailman/listinfo/dataperf























					Reply via email to