RE: How far can U2 scale?
Interesting subject! I think I'm in Brian's camp on this one -- scalability is most dependent on application system and its architecture -- of which the database system is a critical component. I'm wondering where n-tier applications fit into this discussion. I don't think it's a stretch to say that the architecture of most MV applications is at best a 2-tier design... and the client tier tends to be very thin. With such a design, it seems reasonable to say that for a well designed 2-tier application, the performance characteristics and capability of the database system to use available hardware resources are significant factors. What little bit I know about n-tier architecture tells me the database system is a scalability factor, but the addition of other components in the application needed to coordinate application functionality across the various tiers plays a HUGE role. Well designed applications that can scale by adding systems seems like a powerful notion. But, just like the 2-tier application, scalability is still dependent on the capability of the overall application design (including its third-party components) and its capable to use the available hardware resources. N-tier seems like scalability Nirvana to me -- though very difficult to achieve. Are there highly scalable n-tier applications using Universe, Unidata, jBASE, etc? Tom Firl Columbia Ultimate > -Original Message- > From: Dawn M. Wolthuis [mailto:[EMAIL PROTECTED] > Sent: Friday, April 23, 2004 6:50 AM > To: [EMAIL PROTECTED] > Subject: How far can U2 scale? > > > At what point in the life of application software would it be > so large that > you could not (or would not want to) support it with your > existing UniData > or UniVerse database? > > Is there a point where you would be better served by DB2 or > Oracle, for > example due to the scale you are working with? > > I hear people talk about moving way from U2 in order to do > ODBC and use > standard industry tools (and most find that the grass is not > greener for > those purposes), but I don't hear about switching because of > running into > scaling issues. However, we sometimes think of PICK as addressing > small-to-mid size businesses and RDBMS folks sometimes think of their > products as scaling the best. > > So, what's the cut-off for U2? Thanks. --dawn > > Dawn M. Wolthuis > Tincat Group, Inc. > www.tincat-group.com > > Take and give some delight today. > > > > -- > u2-users mailing list > [EMAIL PROTECTED] > http://www.oliver.com/mailman/listinfo/u2-users > -- u2-users mailing list [EMAIL PROTECTED] http://www.oliver.com/mailman/listinfo/u2-users
RE: [UV] ODBC Account Flavor
> My only .02 is that the fully compliant schema seems to > interact better with > odbc SQL clients - probably because using CREATE-TABLE > commands makes the > dictionaries exactly what they should be, and the creation of > things like > UV_TABLES. > We probably won't go so far as to define a SQL schema on the account, but I can definitely see how it removes many of the difficulties in getting and ODBC interface up and running. Thanks for your response... the level of feedback I'm looking for. Tom Firl Columbia Ultimate -- u2-users mailing list [EMAIL PROTECTED] http://www.oliver.com/mailman/listinfo/u2-users
[UV] ODBC Account Flavor
I'm plowing through a project to setup a standard UV-ODBC table and column definition for our application. Our application database accounts are setup as PICK flavored accounts. For ODBC, we're going to setup a separate account for each application account with file pointers to the DATA portion of files and cleaned-up local dictionary files. I've been told by IBM that the "ODBC account must be IDEAL flavor". I know that isn't totally true as I have a proof-of-concept system setup where the ODBC account is setup as PICK flavor. Like application accounts, is the flavor of the ODBC account simply a matter of preference? Or are there "limitations" with ODBC and PICK flavored accounts that I should be aware of? Tom Firl Columbia Ultimate -- u2-users mailing list [EMAIL PROTECTED] http://www.oliver.com/mailman/listinfo/u2-users
RE: Jbase handles multivalue on RDBMS
> > And for the not so faint of heart, jBASE provides the capability to roll > > your own driver(s) to transform your MV database definition into the relational > > database without using the jEDI development kit. > > > > Tom Firl > > Columbia Ultimate > > And I can use my shoe as a hammer, but of course it doesn't work so well that > way. The drivers can be written in BASIC... so any programmer can do it. But, in the context of using DB2 (or some other relational database) as the application's primary data store, writing those drivers in C or C++ will probably yield a little better throughput. You also don't get the other bells and whistles included in the jEDI development kit. Tom Firl Columbia Ultimate -- u2-users mailing list [EMAIL PROTECTED] http://www.oliver.com/mailman/listinfo/u2-users
RE: Jbase handles multivalue on RDBMS
And for the not so faint of heart, jBASE provides the capability to roll your own driver(s) to transform your MV database definition into the relational database without using the jEDI development kit. Tom Firl Columbia Ultimate > -Original Message- > From: djordan [mailto:[EMAIL PROTECTED] > Sent: Wednesday, April 14, 2004 10:27 PM > To: 'U2 Users Discussion List' > Subject: Jbase handles multivalue on RDBMS > > > Hi Steve > > Just to correct you, jbase does not require you to move to > 1NF files to > run on an RDBMS. Jbase will port multi dimensional data across to an > RDBMS and automatically handle the conversion to multiple tables > invisible to the application. The issue is in the quality of the > dictionary, like lengths and data types that RDBMS do not handle > breaking the rules. Jbase does handle a lot of these issues > and I would > assume IBM will incorporate that in U2. Also in such an > environment you > would not move all your files over to an RDBMS, it would make sense to > leave work files and control files in Universe which are usualy the > worst offenders. If you wish to make your application portable in a > future environment like this, look at SQLising your files including > multivalues and starting cleaning your data as this will be > your biggest > issue, not multivalues. > > Just another point, jbase does the same for Cache, which is another > multi-dimensional database, although not PICK. > > Regards > > David Jordan > > -Original Message- > From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] > On Behalf Of Steve Mayo > Sent: Thursday, 15 April 2004 4:01 AM > To: U2 Users Discussion List > Subject: RE: The future of U2 > > > The way that jBase handles the problem is by requiring the database be > flattened out (i.e., no multivalues) and strict data typing. > This is of > course the standard with 1NF databases. Unfortunately for > most of us, it > means a complete redesign of the existing mv database structure. Over > the past several years, all new systems that I have developed > have used > 1NF. Still most of the data still uses multivalues and would > take years > to convert. :-) > > Steve > > -- > u2-users mailing list > [EMAIL PROTECTED] > http://www.oliver.com/mailman/listinfo/u2-users > -- u2-users mailing list [EMAIL PROTECTED] http://www.oliver.com/mailman/listinfo/u2-users
RE: The future of U2
> > U2 TO DB2 ---> Best thing to Happen. H... I don't think I'll touch that one other than to say that only time will tell. Tom Firl Columbia Ultimate -- u2-users mailing list [EMAIL PROTECTED] http://www.oliver.com/mailman/listinfo/u2-users
RE: The future of U2
Has any one heard any specifics about the implementation? I'd be interested in knowing whether or not Universe applications using DB2 as a data store will require setting up a Universe SQL schema. I'm supposing that it will... Tom Firl Columbia Ultimate > -Original Message- > From: Roger Glenfield [mailto:[EMAIL PROTECTED] > Sent: Tuesday, April 13, 2004 8:21 PM > To: U2 Users Discussion List > Subject: RE: The future of U2 > > > I believe the wording was DB2 and then others based on 'demand'. > > Roger > > > -Original Message- > > From: [EMAIL PROTECTED] > [mailto:[EMAIL PROTECTED] > > Behalf Of Ross Ferris > > Sent: Tuesday, April 13, 2004 11:06 PM > > To: U2 Users Discussion List > > Subject: RE: The future of U2 > > > > > > I'd also think that rather than "any" database, the target would > > be DB2 :-) > > > > Ross Ferris > > Stamina Software > > Visage – an Evolution in Software Development > > > > > > http://www.oliver.com/mailman/listinfo/u2-users > > > -- > u2-users mailing list > [EMAIL PROTECTED] > http://www.oliver.com/mailman/listinfo/u2-users > -- u2-users mailing list [EMAIL PROTECTED] http://www.oliver.com/mailman/listinfo/u2-users
RE: mvBase to Pick to UniVerse convertion
> > I have a mvBase VTF file that I would like to restore into UniVerse. > I can get it converted to a Pick "account save" but I don't > have a tape drive, can I mount a cd as a tape drive in UniVerse? > Not really, but you don't need to attach to a tape in order to restore an account from a file. I haven't worked with VTF files, but generally you can do the following to restore the account from a file: UNIX cd /parent/directory/of/account /path/to/uvhome/bin/acct.restore -a /path/to/file.vtf -n -19 Now, create the account in the UV menus or using UniAdmin. Windoze --- cd \parent\directory\of\account type \path\to\file.vtf | \path\to\uvhome\bin\acct.restore.exe - -n -19 Now, create the account in UniAdmin. Note, the -n option indicates the modulo of the files in the account save archive are based on a 2K frame. The -19 option says to convert DC type files to type 19 files (the "file" is a directory and the programs are files within the directory). The -a option lets you specify the file containing the account-save archive. On Windoze, I use the "type" command to read the file and tell acct.restore.exe to read from standard input (the - option) instead of use the -a option -- I've found "acct.restore.exe -a" to be a bit flaky on Windoze. For both platforms, there are other issues such as checking/setting permissions on the account that you need to address. But, generally the steps above can be used to handle the account-restore. Tom Firl Columbia Ultimate -- u2-users mailing list [EMAIL PROTECTED] http://www.oliver.com/mailman/listinfo/u2-users
RE: UV - is there a coldstart equivalent
Universe does not have a "coldstart" process within the database environment. The .rc scripts on UNIX have been mentioned. On Windows, the Resource Kit has a service that can be installed called "AUTOEXNT.EXE". The AUTOEXNT Service allows you to start a custom batch file -- AUTOEXNT.BAT -- when you start a computer without having to log onto the computer on which it will run. I haven't used AUTOEXNT to fire off processes within Universe, but I suppose the AUTOEXNT.BAT script would need to wait for Universe to start up, then fire up your Universe phantom. Tom Firl Columbia Ultimate > -Original Message- > From: Donald Kibbey [mailto:[EMAIL PROTECTED] > Sent: Tuesday, March 09, 2004 9:32 AM > To: [EMAIL PROTECTED] > Subject: Re: UV - is there a coldstart equivalent > > > On a Unix based system you can use the cron facilities to > perform periodic mantenance and the startup script to perform > onetime chores when the UniVerse daemons are first started. > > The same thing can be done under Windoze by using either the > builtin scheduler or a cron substitute. Replacing the > shortcut or start menu link with a script or batch file would > take care of the startup process. > > >>> [EMAIL PROTECTED] 03/09/04 12:04PM >>> > In D3 there is a coldstart process that runs when the > database is started > which is useful from an applications stand point by clearing > status files, > starting application phantoms and general cleanup. > > There does not seem to be an equivalent mechanism in Universe. > > What have people done to provide similar and reliable functionality? > > -Troy > -- > u2-users mailing list > [EMAIL PROTECTED] > http://www.oliver.com/mailman/listinfo/u2-users > > -- > u2-users mailing list > [EMAIL PROTECTED] > http://www.oliver.com/mailman/listinfo/u2-users > -- u2-users mailing list [EMAIL PROTECTED] http://www.oliver.com/mailman/listinfo/u2-users
RE: UniVerse on NT vs *nix - Higher User Counts with W2003?
Simon, I know of at least one site running over 500 users on a W2K box, though I tend to think this is exceptionally high. Given the architecture and limitations of Windows, making any assurance for scalability inside the box WITHOUT TESTING is difficult when you get above 300 users. Tom Firl Columbia Ultimate > -Original Message- > From: Kirkham, Simon [mailto:[EMAIL PROTECTED] > Sent: Tuesday, March 09, 2004 2:49 AM > To: U2 Users Discussion List > Subject: RE: UniVerse on NT vs *nix - Higher User Counts with W2003? > > > Now I'm worried... > > We've got a 300 user enterprise installation runnning on > win2k (enterprise), and the user count regularly goes well > over 300 (388 right now). What symptoms of hitting the around > 300 user ceiling might I see? > > For info, hardware is an 8xcpu dell box, 2Gb ram, 80gb > external raid box. Users are mostly uniobject & telnet > clients, no significant batch jobs during main office hours. > Performance is no problem (except when a redback process goes > nuts, but thats another story). > > Simon > > Simon Kirkham > IS Manager > Taunton Deane B.C. > Tel 01823 356396 > Fax 01823 356329 > > -Original Message- > From: [EMAIL PROTECTED] > [mailto:[EMAIL PROTECTED] > Behalf Of Stephen O'Neal > Sent: 09 March 2004 09:16 > To: [EMAIL PROTECTED] > Subject: RE: UniVerse on NT vs *nix - Higher User Counts with W2003? > > > > > > > > Yes, currently, most W2000 installations max out at about 300 users. > > We may be at a point of transition. The old ceiling of > around 300 users on > W2000 may increase to a higher threshold with W2003. > > Initial tests (although limited) on W2003 point to some of > the inherent > bottlenecks may be removed that were in W2000. > > We have not performed any formal benchmarks (nor do we have > any scheduled > soon) to attempt to scale a W2003 system to maximum user load with an > application. > > (Humor here...) Given the huge sucking sound that comes from > the black hole > in the "UniVerse" called Microsoft, I believe there will be > someone out > there that will upgrade from W2000 to W2003. They may observe more > scalability of their applications as we have in our limited > testing. This > may be the first place where we observe the user counts going > up on W2003 > systems. > > Has anyone performed an upgrade from W2000 to W2003 and > gotten more users > on their system? (Quantify if possible.) > > Curious, >Steve > >Stephen M. O'Neal, CDP >IBM Data Management Solutions >U2 Professional Services, Special Projects) > > > > > > > "Ken Wallis" > > > <[EMAIL PROTECTED]To: "'U2 > Users Discussion List'" <[EMAIL PROTECTED]> > > st.com> cc: > > > Sent by: Subject: RE: > UniVerse on NT vs *nix > > u2-users-bounces@ > > > oliver.com > > > > > > > > > 03/08/2004 06:42 > > > PM > > > Please respond to > > > U2 Users
RE: UV - Database backup
You may not have SUSPEND.FILES as the following link indicates this feature was added on the 9.5C release: http://www.indexinfocus.com/dl/u2list/200211/40606.html I wouldn't be surprised if SUSPEND.FILES didn't make it into the User Ref documentation until 9.6, or later. You might check the UV account VOC file for an item called SUSPEND.FILES... if you have this item, then I would think the SUSPEND.FILES command is available for you to use. Tom Firl Columbia Ultimate > -Original Message- > From: Robert Porter [mailto:[EMAIL PROTECTED] > Sent: Monday, March 08, 2004 10:41 AM > To: [EMAIL PROTECTED] > Subject: Re: UV - Database backup > > > I just searched the admin guide, and the user reference, and > "SUSPEND.FILES" wasn't found in either of them. Could this > be version related? > > Tried it in the development account, and got this: > :SUSPEND.FILES > > Verb "SUSPEND.FILES" is not in your VOC. > > Robert > > >>> [EMAIL PROTECTED] 03/08/04 11:58AM >>> > > SUSPEND.FILES is documented in the UniVerse User Reference manual (and > briefly in the admin manual). The commands are SUSPEND.FILES ON and > SUSPEND.FILES OFF. If you're doing it from the O/S level, > you may want toQ > use the uv -admin -L option, which performs the same thing as > SUSPEND.FILES > ON. uv -admin -U will turn it off. > -- > u2-users mailing list > [EMAIL PROTECTED] > http://www.oliver.com/mailman/listinfo/u2-users > -- u2-users mailing list [EMAIL PROTECTED] http://www.oliver.com/mailman/listinfo/u2-users
RE: How to Add Triggers on Dicts
> > ED VOC DICT.IN.MASTER > 0001: F > 0002: E:\accounts\_source.DEV\D_IN.MASTER > 0003: E:\accounts\_source.DEV\D_IN.MASTER > Conceptually, I don't see anything wrong with what you are doing. This VOC pointer seems to be all right. The work I was on UNIX running 10.1... I don't have a Windows system handy at the moment, so I can't give you're scenario a try. > Ran this > CREATE TRIGGER AUDIT_DICT AFTER INSERT OR UPDATE OR DELETE > ON > DICT.IN.MASTER FOR EACH ROW CALLING 'RC.DATA.TRIGGER Looks good. > > Adding trigger "AUDIT_DICT" > UniVerse/SQL: Unable to open > "E:\accounts\_source.DEV\D_IN.MASTER/DICT.IN.MASTER". > UniVerse/SQL: Could not add trigger "AUDIT_DICT". No idea why the system is trying to open "E:\accounts\_source.DEV\D_IN.MASTER/DICT.IN.MASTER". The error is correct in stating that this path does not exists. But, I have no clue why the system would concatenate the VOC ITEM name -- DICT.IN.MASTER -- to the path name in attribute 2 -- E:\accounts\_source.DEV\D_IN.MASTER -- to derive the path of the file. That does not seem right. I'd probably try setting up a scenario where the values in attribute 2 and 3 are simply the name of the file -- D_IN.MASTER. Otherwise, you might give IBM a call to see if they can help. Tom Firl -- u2-users mailing list [EMAIL PROTECTED] http://www.oliver.com/mailman/listinfo/u2-users
RE: How to Add Triggers on Dicts
I successfully used this technique a few months back during a little troubleshooting exercise... I simply created an F-pointer that specified the D_FileName in both attributes 2 and 3. Essentially, you end up with an entity that gives you similar behavior to the reflexive Q-pointers you see on the traditional PICK platforms. Tom Firl Columbia Ultimate -Original Message- From: Daly, Mark [mailto:[EMAIL PROTECTED] Sent: Tue 3/2/2004 6:28 AM To: 'U2 Users Discussion List' Cc: Subject:RE: How to Add Triggers on Dicts Well.. It wouldn't surprise me if the CREATE TRIGGER command doesn't recognize the 'DICT' keyword. Triggers generally deal with data updates. I guess you could create a dummy file pointer that points to the dictionary as though it were a data file. Then reference that pointer when creating the trigger. BUT - I haven't tried it. Not sure I would try it. Good luck! -Original Message- From: Dennis Bartlett [mailto:[EMAIL PROTECTED] Sent: Tuesday, March 02, 2004 9:20 AM To: 'U2 Users Discussion List' Subject: RE: How to Add Triggers on Dicts The trigger is currently working fine on DATA files. It's just the adding of it to DICT files that's boggling me at present. Once I've got that right, I'll have to think of some way to monitor Type 1/19 files, but that's another day. -- u2-users mailing list [EMAIL PROTECTED] http://www.oliver.com/mailman/listinfo/u2-users -- u2-users mailing list [EMAIL PROTECTED] http://www.oliver.com/mailman/listinfo/u2-users
RE: UniVerse and Backup Exec
I wish I could take the credit... some nice folks at jBASE tracked this information down for me a few years back and I've found it useful for troubleshooting telnet connection issues on Windows. Tom Firl Columbia Ultimate > -Original Message- > From: Gwen Buck [mailto:[EMAIL PROTECTED] > Sent: Friday, February 20, 2004 10:47 AM > To: U2 Users Discussion List > Subject: RE: UniVerse and Backup Exec > > > > > Excellent write up Tom. This is a keeper email. > > Thanks, > > -Doug > > > Yes, Tom. The Windows SharedSection registry tweak write-up > was great. > I apologize for neglecting to acknowledge it sooner! > > Thank you, > Gwen Buck > -- > u2-users mailing list > [EMAIL PROTECTED] > http://www.oliver.com/mailman/listinfo/u2-users > -- u2-users mailing list [EMAIL PROTECTED] http://www.oliver.com/mailman/listinfo/u2-users
RE: UniVerse and Backup Exec
> Dennis also mentioned a registry tweak. I'm guessing that > possibly this is contained in the tech tip that Steve posted. I didn't catch the nature of the errors, but the registry tweak may have to do with the Windows SharedSection parameter that relates to telnet service. Users that connect via telnet attach to a non-interactive desktop in Windows. Each user requires some desktop heap -- a finite resource. You can set the amount of desktop heap allocated to non-interactive desktop by tweaking the SharedSection parameter for Windows. The Windows SharedSection parameter is found in the registry at "HKEY_LOCAL_MACHINE/System/CurentControlSet/Control/SessionManager/SubSystem/Windows" The DWORD value for the Windows key will look something like this: "%SystemRoot%\system32\csrss.exe ObjectDirectory= \Windows SharedSection=1024,3072,512 Windows=On SubSystemType=Windows ServerDll=basesrv,1ServerDll=winsrv:UserServerDllInitializat ion,3ServerDll=winsrv:ConServerDllInitialization,2 ProfileControl=OffMaxRequestThreads=16" The SharedSection ,, string is the parameter of interest. The values of , , and are defined as follows: is the size of the system heap in kbytes - is the size of the interactive desktop heap in kbytes -- that is, each desktop that is created is allocated an amount of kbytes of heap space. - is the size of the non-interactive desktop heap in kbytes -- again, each non-interactive desktop that is created is allocated an amount of kbytes of heap space. The non-interactive heap setting -- -- is the one that affect telnet. The default setting of when the OS is installed is blank. The default value when is blank is 3072. Note, on a system-wide basis there is a finite limit of 48MB of shared heap space. When B/E is installed, the value of is set to 512. This reduces the size of the heap for each non-interactive desktop that is created -- including the non-interactive desktop used by the UV Telnet service. The smaller the heap space of the non-interactive desktop for the UV telnet service, the fewer the number of connections that the service will support. I would expect that if limited heap space is causing the errors in the registry, you would also be having some problems getting users connected, or having users abnormally disconnected. If this is the case, you might increase the size of the heap space allocated to non-interactive desktops () to a value of 1024. You will need to reboot the system after making the change in order for them to be effective. Note, because there is a finite limit of system-wide heap space, increasing the values of and can cause other problems (such are some services not starting). How much system-wide heap space that is used is determined by the applications that create desktops and the size of those desktop. Typically, non-interactive desktops are created by services and an interactive desktop created for the console user. Note, the finite heap space limit of 48MB is a BIG reason why it is difficult to achieve high levels of user scalability for "terminal server-like" applications (such as some of the green screen software that runs on Universe and jBASE) on Windows servers. Tom Firl Columbia Ultimate -- u2-users mailing list [EMAIL PROTECTED] http://www.oliver.com/mailman/listinfo/u2-users
RE: Real Time Data Warehouse
> Another possible name for a real-time data warehouse is > Operational Data Store (ODS). I'm somewhat familiar with the concept ODS, I don't think it will play a role in this project, but it is on my radar. > Any approach to actually porting data to SQL Server, for > example, sounds so > small and innocent until two years down the line you add up > the costs of > hardware, software, training for users and IT, on-going > support, etc and > find that it was a much bigger expense than anyone estimated up front. > I hear what you are saying... on more than one level. Thanks for your reply Dawn. Tom Firl -- u2-users mailing list [EMAIL PROTECTED] http://www.oliver.com/mailman/listinfo/u2-users
RE: Real Time Data Warehouse
> > Just for some more background here is the real reason > you are having to go through the ETL is so that the users can > "play" with your data using 'standard' BI tools like Cognos > against the SQL database ? > Yes. > Also, what USE is the information going to be put to ?! To be determined... but I get your message. Tactical analysis is a critical component in the businesses we cater to. Thanks for your response. Tom Firl -- u2-users mailing list [EMAIL PROTECTED] http://www.oliver.com/mailman/listinfo/u2-users
RE: Real Time Data Warehouse
> I would question how real time the OLAP & BI tools are never mind the > database. Yes, the "real-time" requirement is ambiguous. It will be addressed as we flesh out the tactical analysis requirements for the BI tools. > I would suspect they are looking for a dashboard solution > rather than an OLAP tool. BI, OLAP, and reporting is required by the project (among some other things). BI will be used for tactical analysis using "real-time" data. OLAP will be used for strategic analysis using "point-in-time" data. And, reporting is essentially for generating external reports for customers. > I have been putting a white paper together to try an identify > that PICK > is the platform of choice for this time of environment. > I'm not going to touch that one... though I understand your point ~8^) Thanks for your response! Tom Firl -- u2-users mailing list [EMAIL PROTECTED] http://www.oliver.com/mailman/listinfo/u2-users
Real Time Data Warehouse
So, I'm moving on from my "encrypted database" problem (that was put on hold) and now I have a new, interesting problem. I'm looking at a proposal that seems to demand a solution that is a cross between a "data replication system" and a "data warehouse". The system needs to be able to "Extract" data from a feed up updates to specified Universe or jBASE files in "real-time" (once a minute, or so will suffice), do some "Transformation" on the data, then "Load" the data into a DB2 or SQL Server (not my implementation requirement... don't yell at me). During peak times, I'm supposing could be over a thousand updates per minute written to the data replication feed. I don't know if it's reasonable to expect this system to be able to handle that kind of throughput... that is to be determined. The rationale for the system is to allow people to use standard reporting, OLAP, and BI tools. In industry parlance, I think such a system is called a "Real Time Data Warehouse (RTDW)". So, here's where you can help... I'm brainstorming for design/implementation ideas. First, I'm trying to get the lay-of-the-land of tools and companies that can help with the "ETL" (Extract-Transform-Load) part of this project (is this what DataStage does?). Where do I look? Second, I'm searching for clever ideas about how to create and extract the data feed containing file updates -- such as leveraging UV-DR. I'd prefer to create the data replication feed in isolation from the ETL tool. Seeing as I'm a little lazy (and hoping we won't have to roll our own) I'd like to evaluate off-the-shelf solutions. TIA, Tom Firl Columbia Ultimate -- u2-users mailing list [EMAIL PROTECTED] http://www.oliver.com/mailman/listinfo/u2-users
RE: mkdbfile: create file in another account
Nice catch, Karl. My jBASE-thinking brain completely spaced off the need for F pointers on Universe. It's good to know you're still willing to correct my mistakes ~8^) Tom Firl Columbia Ultimate > -Original Message- > From: Karl L Pearson [mailto:[EMAIL PROTECTED] > Sent: Wednesday, January 28, 2004 10:33 AM > To: U2 Users Discussion List > Subject: Re: mkdbfile: create file in another account > > > That syntax worked for me (sh -c on Linux rather than DOC /C) but, it > doesn't make a DICT file, so the file isn't accessible through TCL > commands. I had to add a DICT to the VOC entry I also had to > add as the > command doesn't create either a DICT file or the VOC entry... I repeat > myself. > > Karl > > > On Tue, 2004-01-27 at 19:31, Kate Stanton wrote: > > MessageThanks. That (using DOS \C rather than sh -c) got > rid of the error message, but it did not seem to create a file. > > > > Looks like I have the wrong syntax, but it is rather hard > to look up the documentation! > > > > I still feel uncomfortable using an undocumented feature - > no security it will not disappear or change. > > > > - Original Message - > > From: Daly, Mark > > To: 'U2 Users Discussion List' > > Sent: Wednesday, January 28, 2004 10:23 AM > > Subject: RE: mkdbfile: create file in another account > > > > > > Well, I just posed this question - and mkdbfile was the > suggestion that won! > > > > However, I didn't create a VOC entry. The idea (I > believe) is to execute this command at the OS prompt - not TCL. > > > > Since it looks like you're on Windows it would be > something like so: > > > > EXECUTE 'sh -c "cd C:\BeaconUV\DevelData\XXINV\KKINV; > ':SYSTEM(32):'\bin\mkdbfile DATA \XXINV\KKINV 30 1 4 20 50 80 1068' > > > >It's not documented, since it's really an internal UV > command - as opposed to a TCL statement of BASIC function. > > > > The CREATE.FILE verb needs to be enhanced to perform this > function - and prevent us from dabbling in the 'bin' > directory. But until then - this is the way to go. > > > > HTH, > > > > Mark. > > > >Original Message- > > From: Kate Stanton [mailto:[EMAIL PROTECTED] > > Sent: Tuesday, January 27, 2004 4:17 PM > > To: [EMAIL PROTECTED] > > Subject: mkdbfile: create file in another account > > > > > > Does anyone know anything about using mkdbfile (from UV > bin) to create a file in an account other than the current > user account? > > > > A colleague told me about it, but it does not seem to be > documented, and does not seem to work on my > > > >UV 10.0.17 system on Windows 2000, as it does on his UV > 9.5.2.1 on Windows something. > > > > I tried, according to his instructions: > > VOC mkdbfile > > 0001: V > > 0002: mkdbfile > > 0003: E > > 0004: FG > > 0005: > > 0006: PICK.FORMAT > > > > Then, from TCL: mkdbfile > C:\BeaconUV\DevelData\XXINV\KKINV 30 1 4 20 50 80 1068 > > where: > > C:\BeaconUV\DevelData\XXINV\KKINV is pathname of file to create > > 30 is file type > > 1 is modulo > > 4 is separation > > 20 is hash type > > 50 is max load > > 80 is split load > > 1096 is large record size > > > > This gave error message: invalid filetype specified > > > > I feel very nervous about using something that is not > documented, so presumably may not be reliably supported. > > > > Anyone know anything, please? > > > > Cheers, Kate > > > > Kate Stanton > > Walstan Systems Ltd > > 4 Kelmarna Ave, Herne Bay, Auckland, New Zealand > > ph +64 9 360 5310 fax +64 9 376 0750 > > ah +64 9 378 9594 > > [EMAIL PROTECTED] > > > > > > > > > -- > > > > > > > ___ > > u2-users mailing list > > [EMAIL PROTECTED] > > http://www.oliver.com/mailman/listinfo/u2-users > > > > > > > ___ > > u2-users mailing list > > [EMAIL PROTECTED] > > http://www.oliver.com/mailman/listinfo/u2-users > -- > Karl L. Pearson > Director of IT, > ATS Industrial Supply > Direct: 801-978-4429 > Toll-free: 888-972-3182 x29 > Fax: 801-972-3888 > http://www.atsindustrial.com > [EMAIL PROTECTED] > > > ___ > u2-users mailing list > [EMAIL PROTECTED] > http://www.oliver.com/mailman/listinfo/u2-users > ___ u2-users mailing list [EMAIL PROTECTED] http://www.oliver.com/mailman/listinfo/u2-users
RE: mkdbfile: create file in another account
Title: Message I don't think you need to "cd" command as the mkdbfile will accept the absolute path to the file you are creating. If you need to execute multiple commands in the same DOS shell, then separate the commands with an ampersand (&) instead of semi-colon (;). I'd be a little surprised if mkdbfile disappeared or required changes to your code, but you never know. I'm pretty sure the Universe installation scripts/programs -- among other things -- use mkdbfile quite extensively. At any rate, give the following a whirl... EXECUTE 'DOS /C "':SYSTEM(32):'\bin\mkdbfile C:\BeaconUV\DevelData\XXINV\KKINV 30 1 4 20 50 80 1068"' Tom Firl Columbia Ultimate -Original Message-From: Kate Stanton [mailto:[EMAIL PROTECTED]Sent: Tuesday, January 27, 2004 6:32 PMTo: U2 Users Discussion ListSubject: Re: mkdbfile: create file in another account Thanks. That (using DOS \C rather than sh -c) got rid of the error message, but it did not seem to create a file. Looks like I have the wrong syntax, but it is rather hard to look up the documentation! I still feel uncomfortable using an undocumented feature - no security it will not disappear or change. - Original Message - From: Daly, Mark To: 'U2 Users Discussion List' Sent: Wednesday, January 28, 2004 10:23 AM Subject: RE: mkdbfile: create file in another account Well, I just posed this question - and mkdbfile was the suggestion that won! However, I didn't create a VOC entry. The idea (I believe) is to execute this command at the OS prompt - not TCL. Since it looks like you're on Windows it would be something like so: EXECUTE 'sh -c "cd C:\BeaconUV\DevelData\XXINV\KKINV; ':SYSTEM(32):'\bin\mkdbfile DATA \XXINV\KKINV 30 1 4 20 50 80 1068' It's not documented, since it's really an internal UV command - as opposed to a TCL statement of BASIC function. The CREATE.FILE verb needs to be enhanced to perform this function - and prevent us from dabbling in the 'bin' directory. But until then - this is the way to go. HTH, Mark. Original Message-From: Kate Stanton [mailto:[EMAIL PROTECTED] Sent: Tuesday, January 27, 2004 4:17 PMTo: [EMAIL PROTECTED]Subject: mkdbfile: create file in another account Does anyone know anything about using mkdbfile (from UV bin) to create a file in an account other than the current user account? A colleague told me about it, but it does not seem to be documented, and does not seem to work on my UV 10.0.17 system on Windows 2000, as it does on his UV 9.5.2.1 on Windows something. I tried, according to his instructions: VOC mkdbfile 0001: V 0002: mkdbfile 0003: E 0004: FG 0005: 0006: PICK.FORMAT Then, from TCL: mkdbfile C:\BeaconUV\DevelData\XXINV\KKINV 30 1 4 20 50 80 1068 where: C:\BeaconUV\DevelData\XXINV\KKINV is pathname of file to create 30 is file type 1 is modulo 4 is separation 20 is hash type 50 is max load 80 is split load 1096 is large record size This gave error message: invalid filetype specified I feel very nervous about using something that is not documented, so presumably may not be reliably supported. Anyone know anything, please? Cheers, Kate Kate StantonWalstan Systems Ltd4 Kelmarna Ave, Herne Bay, Auckland, New Zealandph +64 9 360 5310 fax +64 9 376 0750ah +64 9 378 9594[EMAIL PROTECTED] ___u2-users mailing list[EMAIL PROTECTED]http://www.oliver.com/mailman/listinfo/u2-users ___ u2-users mailing list [EMAIL PROTECTED] http://www.oliver.com/mailman/listinfo/u2-users
RE: mkdbfile: create file in another account
Title: Message ...or DOS /C if you are on Windows. -Original Message-From: Logan, David (SST - Adelaide) [mailto:[EMAIL PROTECTED]Sent: Tuesday, January 27, 2004 1:39 PMTo: U2 Users Discussion ListSubject: RE: mkdbfile: create file in another account Hi Kate, These commands are designed to be run from the unix or (blerk, I hate this word) dos environment. Running them from TCL would require an sh -c type command. Regards David Logan Database Administrator HP Managed Services 139 Frome Street, Adelaide 5000 Australia +61 8 8408 4273 +61 417 268 665 -Original Message-From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Kate StantonSent: Wednesday, 28 January 2004 7:47 AMTo: [EMAIL PROTECTED]Subject: mkdbfile: create file in another account Does anyone know anything about using mkdbfile (from UV bin) to create a file in an account other than the current user account? A colleague told me about it, but it does not seem to be documented, and does not seem to work on my UV 10.0.17 system on Windows 2000, as it does on his UV 9.5.2.1 on Windows something. I tried, according to his instructions: VOC mkdbfile 0001: V 0002: mkdbfile 0003: E 0004: FG 0005: 0006: PICK.FORMAT Then, from TCL: mkdbfile C:\BeaconUV\DevelData\XXINV\KKINV 30 1 4 20 50 80 1068 where: C:\BeaconUV\DevelData\XXINV\KKINV is pathname of file to create 30 is file type 1 is modulo 4 is separation 20 is hash type 50 is max load 80 is split load 1096 is large record size This gave error message: invalid filetype specified I feel very nervous about using something that is not documented, so presumably may not be reliably supported. Anyone know anything, please? Cheers, Kate Kate StantonWalstan Systems Ltd4 Kelmarna Ave, Herne Bay, Auckland, New Zealandph +64 9 360 5310 fax +64 9 376 0750ah +64 9 378 9594[EMAIL PROTECTED] ___ u2-users mailing list [EMAIL PROTECTED] http://www.oliver.com/mailman/listinfo/u2-users
RE: Database Level Encryption
Yes, using a cryptographic file system to provide encryption services is my preferred strategy. But, we're still negotiating requirements so I'm trying to keep my high-level design options open. As JayJay pointed out, there are some "issues" with this solution. Essentially, after OS security is breached your data is exposed. That may, or may not be good enough... depending on your viewpoint. Some people trust their capability to lock down the OS. And other people are pessimistic about their capability to lock down the OS to their satisfaction and they want another barrier between the data and intruder. Tom Firl Columbia Ultimate > -Original Message- > From: djordan [mailto:[EMAIL PROTECTED] > Sent: Friday, January 23, 2004 5:52 PM > To: [EMAIL PROTECTED] > Subject: RE: Database Level Encryption > > > Hi Tom > > I haven't follow the total chain of this, but do you want the data > encrypted for external users or for other universe users. > Could you not > do something like with MS Windows where you set the directory to > encryption, the user does not have to do anything as the OS does the > encryption. You would not have to change anything in the > application or > in Universe, as Windows will encrypt/decrypt the Universe IO to the OS > files. The Datafiles will be encypted. > > Just an idea > > Regards > > David Jordan > Managign Consultant > [EMAIL PROTECTED] > > Dacono Holdings Pty Ltd > Business & Technology Consulting > PO Box 909 > Lane Cove > NSW 2066 > Australia > Ph 61 2 9418 8329 > Fax 61 2 9427 2371 > www.dacono.com.au > > > > -Original Message- > From: Tom Firl [mailto:[EMAIL PROTECTED] > Sent: Friday, 23 January 2004 2:57 AM > To: [EMAIL PROTECTED] > Subject: RE: Database Level Encryption > > > Thanks Wol, I'll let you know what we end up doing for the encryption > part of the project... The requirements for encryption strength still > need to be worked out, but I don't expect we'll be dealing with > ultra-strong encryption. > > The bigger concern at this point is finding a way to slide > encrypt/decrypt operations between the app and the data > store. Neither > Universe nor the application are well suited for this type of design > change. jBASE is much better at dealing with I/O layer > extensions, but > that is not an option for this project. Anyway, it's looking > like we'll > have to create an extensible I/O library for the application, then add > an encrypt/decrypt layer... it's a pretty good design principle, but > making the change on a 20 year-old, monolithic application is > a big job. > > Tom Firl > Columbia Ultimate > > > -Original Message- > > From: Anthony Youngman > [mailto:[EMAIL PROTECTED] > > Sent: Thursday, January 22, 2004 1:19 AM > > To: [EMAIL PROTECTED] > > Subject: RE: Database Level Encryption > > > > > > I've written an enigma encryption function in the past. I'm > not saying > > > it was any good ... and in these circumstances it might well not be. > > > > But it would be very easy, using such a function, to say > "only encrypt > > > certain characters, leave others unchanged", so you could > scramble all > > > characters and numbers but leave uv marks alone. This has other > > repercussions, like giving away field lengths - you may not > want to do > > > that. > > > > And, DO discuss what you plan to do here. There are hundreds > > of examples > > of people thinking up clever schemes, making elementary > mistakes, and > > the resulting "security" is non-existent. Think Adobe/Sklyarov. Some > > clever clogs at Adobe took what was actually a good > encryption system, > > "enhanced" it, and the end result was a Caesar cipher. In > > case you don't > > know, that's "shift every character x places to the right in the > > alphabet" cipher. I cracked loads of those as a schoolkid. > > > > Security is ONLY effective if you know you can tell an attacker > > everything except the key, and you know that that > information is not > > going to help him one little bit. > > > > The other thing, of course, is all this effort and everything > > worth it? > > We sell a product that has time value. It's encryption-protected, of > > course, so all our customers get the same CD but can only access the > > bits they've paid for. But rather than go for ultimate > security, we've > > gone for something that (a) isn't e
