Keeping PK's Unique Across (potentially) Disconnected Sites.
Apr 17, 2008
Hi All,
I'd like to throw this idea 'out there' to see if I'm missing something I'll later regret.
I'm looking to resolve a scalability issue within our point-of-sale program. Currently the PK on transactional tables (sales and orders) is created by the application layer using a 'MAX(PKCol) + 1' mechanism. Obviously this requires that all users of the system, whether they're local or remote, have current data at any time they wish to insert. It's this limitation I'd like to remove. Most sites are using MS SQL Server 2000. No sites use anything specific to a later version.
By having a PK that can be generated independently of a 'master' database we can overcome this issue. The PK values will need to be unique within a 'group' of shops and able to be generated by a program operating at any level. From 'head office' which manages a number of shops, to the server at a given shop and even the register / till itself should be able to create ID's while disconnected from the server (using a local database).
It seems there's three main ways to accomplish this:
- Identities,
- MachineID, CurrentPK composite.
- GUID's
Identities: I've ruled out identities as I believe the administration overhead of dealing with them makes them impractical (there may be several hundred registers and therefore as many ranges to be set up within a group).
MachineID, CurrentPK composite: The MachineID references a Machine table which has an entry for each ethernet MAC address which connects to the database. The reason I chose to store the MAC in another table rather than simply using it as column is that I'm fetching it from sysprocesses.net_address(nchar(12)) and believe it's computationally cheaper to use an int than a text column. This mechanism means that we can still expose the PK to the user in some cases (eg: InvoiceNumber printed on a receipt). When the local database is not up to date (usually due to network problems) there will be cases where the CurrentPK will be duplicated but kept unique since it's coupled with the new MachineID. The big drawback to this method is that all current code will need to be revised to deal with the composite keys (this will be a significant amount of development).
GUIDs: Ugly to look at and time-consuming to type. They're not something which you'd expose to a user unmodified so realistically this means altering existing code to use a new 'user friendly' number where the PK is currently exposed to them. The use of GUIDs rule-out the use of clustered indecies on tables they're the PK for lest most inserts cause a page split. The splits would also necessitate more frequent index defrags / rebuilds. Using a non-clustered index incurs a penalty Vs a non-fragmented clustered one (doesn't it?) so while this avoids page-splits it comes at a cost.
After all that I think the best solution is to use GUIDs with a non-clustered index for each of the PK's. While it might not be the fastest of the options (slower reads/joins Vs composite PK) it will be significantly faster to develop while maintaining acceptable performance.
I have this query and it works except for I am getting duplicate primary keys with unique column value. I want to combine them so that I have one primary key, but keep all the columns. Example:
Key column 1 column 2 column 3 column 4 A 1 1 A 2 2 B 2 3 B 5 5
it should look like:
A 1 1 2 2 B 2 3 5 5
Here is my query:
SELECT * FROM [TLC Inventory].dbo.['2014 new$'] WHERE [TLC Inventory].dbo.['2014 new$'].mis_key LIKE '2%' AND dbo_Product_Info#description NOT LIKE 'NR%' AND dbo_Line_Info#description NOT LIKE 'OBSOLETE%'
I am trying to pass in a query through reporting services where part of the query string is an <A HREF= "Http:// ......... URL string. Before any kind of execution can occur I get the error mentioned in the title.
I have tried to research this and find suggestions if it was my own .net code but how to turn it off in reporting services? The URL is valid and this is a controlled environment so I am not particularly worried about somebody using this as an exploit.
This is a C++ / ADO / SQL question. Maybe not the right forum but I am guessing there are some programmers out there ...
I am trying to use ADO disconnected recordset to insert data into a sql table. I am using AddNew(vField, vValue) with UpdateBatch(). The code below does not throw any exceptions ... but does not add data to the table.
Any comments are appreciated, Thanks, Chris
void CTestApp::TestDatabaseUpdateBatch1a(void) { int nDataCount = 0; long nIndex = 0; long nIndex2 = 0;
try { // When we open the application we will open the ADO connection pConnection.CreateInstance(__uuidof(Connection));
// Replace Data Source value with your server name. bstr_t bstrConnect("Provider='sqloledb';Data Source='SQLDEV';" "Initial Catalog='AlphaNumericData';" "User Id=cmacgowan;Password=cmacgowan");
// Open the ado connection pConnection->Open(bstrConnect,"","",adConnectUnspecified);
// Create an instance of the database pRecordset.CreateInstance(__uuidof(Recordset));
// Select the correct sql string. Note that we are creating an // empty string by doing a select on the primary key. We are only // doing inserts and we do not want to bring data back from the // server
csSQL = "SELECT * FROM dbo.AAMacgowanTest WHERE RecordId IS NULL"; // csSQL = "SELECT * FROM dbo.DICastRaw1Hr";
// Test to see if the recordset is connected if(pRecordset->GetState() != adStateClosed) { // The recordset is connected, we will see if we are // at the end
if((pRecordset->BOF) && (pRecordset->GetadoEOF())) { // The recordset is empty bIsEmpty = false; }
// COleDateTime is a wrapper for VARIANT's DATE type. COleVariant is // a wrapper for VARIANTs themselves. If you need to create a // variant, you can say: COleDateTime oledtCurrentDate2 = COleDateTime::GetCurrentTime();
// Convert the OleDateTime to the varient COleVariant vCurrentDateTime2(oledtCurrentDate2);
//Set the DATE variant data type. memset(&st, 0, sizeof(SYSTEMTIME)); st.wYear = 2000; st.wMonth = 1; st.wDay = 1; st.wHour = 12;
// vect is a vector of COleSafeArrays containing the records for(i = 0; i < 10; i++) {
// Setup the data nValueIndex = 0; vValueList[nValueIndex].vt = VT_BSTR; vValueList[nValueIndex].bstrVal = ::SysAllocString(L"BLUE"); nValueIndex++;
We're constructing a three-tier application. We want the middle tier to extract a recordset from the database, disconnect from the database, then pass the recordset to the front tier. After changes have been made by the front tier, it will pass the recordset back to the middle tier, which will reconnect, and update the database. The problem: Using stored procedures, the recordset is no longer available once the connection has been closed. Is there any way, using stored procedures, to be able to keep the recordset available after the connection has been closed, either by preserving it, copying it, or otherwise?
I've an issue while calling Oracle Stored procedure from MS SQL Server 2012 using LINKED SERVER, It gets connected and do the execution, But sometimes, it was disconnected and says this message 'A severe error occurred on the current command. The result if any should be discarded'.
I'm having a problem with two mirrored databases, using SQL Server 2005 SP2, that autmatically failed over last night for unknown reasons. I was trying to fail them back over to the primary SQL server because it says the mirror is disconnected and out of sync. Other databases failed over too, but I was able to fail those back over without a problem. It's just these two. I removed the mirror from the secondary server thinking that this would allow me to restore the database back on the primary server, but that didn't help.
These two databases show a status of (Principal, Disconnected / In Recovery) and I still cannot do anything with the two databases on the primary server. When I try to pull up the properties for them, I get the following error:
"Database <Name> is enabled for Database Mirroring, but neither the partner nor the witness server instances are available: the database cannot be opened.(Microsoft SQL Server, Error: 955)"
I cannot delete, detach, Alter or do anything with the databases. If I could just delete them that would be fine so I can just do a restore, but I can't seem to do anything.
Does anyone know what I can do to resolve this problem.
naresh writes "I would like to learn sql server 2000 and I do not have any programming experienc at all so how do i learn this programme. Do you have any suggestion or any basic books or materials you guys can refer to me
I have a mobile application written in vb.net using MsSQL and I want to use disconnected datasets to sync up to a remote MySQL database. is this possible? Its a two way sync, i.e I download and upload info.
Its been almost 10 years since I have had to do work with good-old RecordSet objects...
I am filling a RecordSet with data returned from a SQL server via a stored procedure. I then set the ActiveConnection property to Nothing in order to disconnect it so I can make changes to it.
But when I try to set the value on a given row I get back a "Multiple-step operation generated errors. Check each status value" error message. My understanding is that this is indicative of trying to use the wrong datatype. I have verified that the type is correct (I am dealing with integers) so I am at a loss for what the problem could be.
Oddly enough, when I look at the Attribute property of the field I get back a value of 112. When you break it down I think that value indicates the row value is read-only? (Could that be my problem? Just a really bad/unhelpful error message?) But if I try to change it I get back a message saying cannot be done since RecordSet is already open.
I have a .Net database application that we've successfully deployed in a connected environment. Now we have a client that has the need to store data on a central SQL Server and publish that data out to tablet PC's that will be able to disconnect from the central SQL Server. At some point those tablet PC's will come back in and connect to the central server via VPN and will need to push their changes back to the server. Some fundamental questions:
1) Am I correct in assuming that replication is the best way to accomplish this?
If so,
2) Which replication type sounds appropriate to the above scenario?
3) Am I correct in assuming that the tablet PC's will need some version of SQL Server to support editing of the data in the disconnected state?
My perhaps incorrect first take on this was that we could use a licensed SQL Server on the central server and SQL Server Express as a replication subscriber on the tablet PC's.
While a database upgrade schema changes were being made the Mirror became disconnected will this recover itself when it reconnects or will it be the case that we will have to copy the db files to the Mirror and set it up from scratch.
“HELP !! We’ve lost about 25 client’s websites. The databases were backed up along with all the actual files contained within each CSK….in addition, all the original databases are intact & can be reattached to the new SQL server…..the problem that exists where the original CSK files do not recognize the original database once it is reattached to the new SQL server. Any help would be most appreciated. This is the error…… Login failed for user 'DARRYL1ASPNET'. Description: An unhandled exception occurred during the execution of the current web request. Please review the stack trace for more information about the error and where it originated in the code. Exception Details: System.Data.SqlClient.SqlException: Login failed for user 'DARRYL1ASPNET'.Source Error:
An unhandled exception was generated during the execution of the current web request. Information regarding the origin and location of the exception can be identified using the exception stack trace below. Stack Trace:
If I were going to look into investing into a virtual dedicated server and wanted to save LOTS of $$$ by offerring SQLExpress (FREE) vs. SQL Server Enterprise $25,000 sticker price. The Web sites that I build are for small businesses. Is there an issue in using SQLExpress on the Internet? Considering it is the Web application account ID accessing the database so that is one user hitting the DB. I am unclear on how I need to evaluate multiple calls to the system depending on how many users are on the Web site opening the connection performing SQL Commands then closing the connections all with the same user creditials. <moojjoo />
SQLTeam.com SQLSERVERCENTRAL.com TekTips.com MSDNSQL Forum forums.sqlmag.com
I hate to ask for information such as this but with the low response to my questions I was wondering anyone knows of a better site for SQL Server database mail issues?
I am trrying to figure out what the best replication would be to use and or setup...
Her eis the current goal and structure..
We are just moving over to a new custom POS system that will be using SQL databases....We have have three locations and we want each location to be independent in case of network connectively failures to our primary location.
Basically, all three locations will be running SQL server 2005 and the POS app... We want replication to occur overnight, so that each location will have the other locations transactions from the previous days, etc...
Essencially I want all three locations to "syncronize" their data every night....basically two-way replicaiton between all three sites...
Master Site will have say databaseA that the local POS system will use Sencond Site will have say databaseA that the local POS system will use as well
Third Site will have say databaseA that the local POS system will use well...
I have the following code but it keeps erroring on the last line and I'm unsure as to why it is doing it???
Here is the error message
Msg 102, Level 15, State 1, Line 42
Incorrect syntax near '@Sites'.
declare @Sites varchar(50)
declare @Kit_No char(20)
declare @Location char(2)
set @Location = 'Ho'
set @Kit_No = 'mo1k'
if (SELECT sitetype from gss.dbo.kup_regions where region_code = @Location) = 10
begin
set @Sites = '''Pe'',''Hg'',''Vo'',''' + @Location + ''''
end
select
KR.Region_Code,
KR.Region_Name,
Z.Qty,
Z.Kit_Description,
Z.BookedOutToDate as Usage,
Z.Local_Cost,
(select overstock from gss.dbo.vGss_overstock where region_code = Z.region_code and kit_no = 'm01k' )as Rolling_Avg,
C.Symbol,
Z.FOB,
(SELECT
Price
FROM
gss.dbo.FedEx_Rates Fed
WHERE
SourceRegion = (SELECT Region_Code FROM gss.dbo.KUP_Regions WHERE Region_Name = 'penistone')
AND Weight = (SELECT MAX (Weight) FROM gss.dbo.FedEx_Rates WHERE Weight < (SELECT Packed_Unit_Weight From gss.dbo.KUP_Kits WHERE Kit_Code = (select Kit_Code from gss.dbo.GSS_Kits where Kit_No = 'm01k' ))+0.5)
I have a system use MS SQL 2005 & .NET 2.0, my tables don't have rowversion, but I heard SQL 2005 manage a rowversion by itself, can I use this to do a "ConflictDetection".All I try to do is I want to get a error when I try to update a row which been modified by someone else after I read row. Thanks.
set rsSearch = Server.CreateObject("ADODB.Recordset") rsSearch.CursorLocation = adUseClient rsSearch.LockType = adLockBatchOptimistic
sSql = "some sql stmt" 'this works fine on its own
rsSearch.Open sSql, cnTraining
set rsSearch.ActiveConnection = nothing
cnTraining.Close set cnTraining = nothing
%>
My goal is to get a disconnected recordset. The problem here occurs when i try to use the adUseClient value (3) for the CursorLocation. If I don't use a 3 in the CursorLocation it works fine. Hoewever i'm almost certain i have to use the 3 in order to disconnect the recordset. Any ideas? Is my connection string not set up properly to get a disconnected recordset?
We have mirroring set up on SQL 2005(SP2) on windows server 2003 servers for 6 production servers to DR servers. It is high performance mode(asynchonous). We are using fully qualified server names with default 5022 port. Prod and DR are using same domain service account.
We had network outage for about 30 minutes last weekend and after that network was restored back to normal. Also few SQL 2000 servers have logshipping and it automatically started syncing up after network was restored. However mirroring didn't start automatically. Partners were in disconnected state and endpoints were stopped. For one database, log grew to 40 GB using most of the disk space.
Then we maually had to run
ALTER ENDPOINT Mirroring STATE = STARTED
and then it started syncing up.
Now the question is why it doesn't start syncing up automatically after network is back to normal and recognize the partner? Is there anyway we can setup timeout or any parameter like that to specify how many times or how long partners try to connect with no luck and then give up?
We're using ADO disconnected recordsets. On SQL 2000, we could update these on the client (without propagating the changes to the server) even if the underlying view or table was non-updatable.
When running our apps against SQL 2005 (using the same client-side environment), we can no longer change any attributes of those disconnected recordsets, that connect to a non-updatable database object (the rest of the app runs fine, we can update all updatable database objects through disconnected recordsets) . Does SQL 2005 respond to such calls differently from SQL 2000, so that ADO recordsets are built in a new way (which makes them read-only in our setting)?
I have a SQL 2005 server on a separate machine from my IIS machine, and anytime the SQL server restarts (like for last night's automatic updates) the connection pooling seems to break. Among the apps on this IIS box is Community Server 2.1, along with some other custom-built apps. The only way to resolve it is to stop and restart the IIS services. Previously all SQL-dependent apps on that IIS box broke when SQL was reset, but I seemed to have addressed it, at least on my custom-built apps, by forcing a TCP connection in the connection strings (instead of the default named pipes method). I did change the Community Server connection strings as well, but for some reason that didn't seem to work. Has anyone else experienced similar problems with interruptions in SQL connectivity bringing down apps that connect to that SQL server? I hate to turn off automatic updates just to make sure my ASP.NET apps are available. I apologize if this isn't clearly an ASP.NET question, but hopefully someone can help me out. Thanks,Josh
I'm very new at this and found this video very helpful. I downloaded Visual Web Developer 2005 Express Edition and following the direction except for changing a few column names. When I get to the point of viewing in browser I get the drop down box but nothing in it and the table doesn't show up either. I copied all the data to my website and tried to view it there but got some error that I don't understand. Can someone please help. The website I uploaded the files to is ocbeachrentals.net/default.aspx. Thanks.
I am wondering how people maintain their SQL Servers which run at several customers sites and disk space is getting smaller and smaller? I want to say that we have tables in SQL dbs which hold a lot of date consisting of statistics, errors, logs etc. They grow and grow and existing data is not needed anymore as soon as the data get older than let's say for one year. How do you overcome the problem reducing the tables but not charging the system too much as the major application also runs on the same server?
I am new to RDO and SQL I have been using DAO and know that the syntax is similar but the connections to the database is different. Does anyone know of a good site that explains this?
We have a large SQL database, and we need to send out updated records to many clients' sites which are not connected.
We currently have a tool which looks at the audit log of changes we made, creates a file based on this, which is then emailed to our clients. They then run a tool we created to remerge the changes.
I suspect SQL server replication might make all this possible. Am I right? Can SQL server produce a file automatically which can be applied to a remote database to update the tables as appropriate? From looking at some replication stuff it looks to me like you have to have the servers on the same network.
