Optimising Merge Replication For Both Server And Client Subscribers
Jan 4, 2007
I need to merge replicate data to two different types of subscribers:
Clients subscribers which will have a very small percentage of the data from the central database. The data on these machines will be managed using dynamic filtering on host_name()
Server subscribers which will manage a copy of all the data from the central database
There will be far fewer server subscribers than client subscribers.
As I see it I have two options for the configuration
1) Use two separate merge publications one which is filtered and one which isnt
2) Use a single merge publication and setup the filtering so that the server subscribers receive all the rows
Which option is likely to lead to better performance?
With option 1) there would be 2 complete sets of replication metadata which need to be maintained so I am tending towards option 2. Are there any disadvantages in using a dynamic filter to return a very large number of rows?
I am trying to create Replication Topology (Merge Replication) like below.
Subscriber1 --> Publisher <-- Subscriber2.
I have created both subscribers with Subscription Type as Server with Priority as 75. I am updating the Column A of Row_10 in Subscriber1 on time say 11 am. After i am running the Starting synchronizing agent from Subscriber1. The value propagated to Publisher now publisher contains the latest value in Column A. Uptonow the Subscriber2 is not synchronized with Publisher.
Now in Subscriber2 also Column A of Row_10 is updated say at 11.10 am. Actually now Publisher contains the value from Subscriber1 for that Column and in Subscriber2 we have the same column updated.
Now i am running the Synchronization in Subscriber2, i am getting the result which is not expected. Here Publisher's value is propagated to Subscriber2. But as per real scenario Subscriber2 has the Latest value which is updated on 11.10 AM.
I don't know what am i missing here. Actually merge replication should see the time stamp and it has to decide winner. But here it always considers publisher as a winner and puts the data to Subscriber.
I read the BOL on how the publisher will had out identity ranges to subscribers, but it was not clear if this was also the case for anonymous subscribers. Will merge replication with identity columns work with anonymous subscribers that sync via HTTPS?
I posted a more detailed technical question a couple of weeks ago with Subject = "Compact Edition 2005 Subscribers - Merge Agent failed after detecting that retention-based cleanup has deleted metadata" But I got no replies.
Now my question is simpler: Is SQL Server 2005 Merge Replication reliable with Compact Edition subscribers on WindowsXP tablets?
We deleted all users' local databases a couple of weeks ago, and republished two publications with retention = 60 days. Now we have at least one user who can't sync with the server, and replication monitor shows the error below. But if she e-mails her sdf file to a support person, they can successfully sync her database. We're sure that it's not a permissions issue, which is a normal suspicion when one user can perform a task that another can not. It doesn't seem like a retention problem, because she's well inside the retention period. She can successfully sync with the publication with upload/download tablers, but not with the publication with download only tables. Both publications were created on the same day, by the same DBA, with nearly identical properties.
This problem user has operated in the past with no problem, and she's configured identically to all of the others. It just seems like random flakyness with merge replication for Compact Edition subscribers.
I feel stupid saying that, because it sounds like I believe in a magical ghost in the machine. So my question is - IS MERGE REPLICATION A LITTLE BIT FLAKY FOR COMPACT EDITION SUBSCRIBERS ON XP MACHINES?
We've applied SP2 to the server, but no subsequent HotFixes.
We're trying to sell users and management on developing a larger occasionally connected system with the same replication scheme, so we need to know if it deserves our confidence.
Error messages:
The Merge Agent failed after detecting that retention-based metadata cleanup has deleted metadata at the Publisher for changes not yet sent to the Subscriber. You must reinitialize the subscription (without upload). (Source: MSSQL_REPL, Error number: MSSQL_REPL-2147199402) Get help: http://help/MSSQL_REPL-2147199402
I am looking for bidirectional transactional replication using updatable subscribers (queued or immediate) . Is it possible to replicate the image data from the updatable subscribers to the publisher. I understood that the Image data can't be replicated to the publisher from the updatable subscriber. I am not using the WRITETEXT or UPDATETEXT. I am using just INSERT and UPDATE for image data type transactions.
We are running merge replication, SQL Server 2005 Enterprise with SQL Mobile 2005 (Windows Mobile 5) subscribers. Partitions are filtered on HOST_ID.
Ocassionally we experience a situation where a subscriber experiences an unusually long synchronization duration, and upon examining Replication Monitor, it appears that twice the number of rows, or X the number of rows (up to 7 times the number of rows) that should have been inserted are recorded as synchronized for the session: once the normal amount as inserts and once the normal amount as updates. This occurs for all tables in the subscription. This occurs on a first time synchronization to an empty subscriber database where there should be only inserts taking place.
I have examined the snapshot partition folders for these users on the file system and they appear to be identical in size and content as other subscribers. Checking the last partition snapshot job run and other characteristics for the the subscriber in question, everything appears to be the same as other subscribers functioning normally.
The HOST_ID for us is an employee ID used to filter employee specific data. I have seen this happen if the subscriber changes the value for the HOST_ID used in filtering, after the mobile database has already been populated (2 employees attempt to use the same mobile device and database). But, we have seen this happen recently where the HOST_ID was apparantly never changed midstream.
This just started happening recently. The only modification around the same time frame was the implementation of a custom business logic handler/custom conflict resolver that performs like "Latest Wins" but has logic added to update the a last-update datetime column for selected transaction tables at time of synchronization, so that an SSIS job can detect the changed rows for copying incremental database changes to another application database. This all seems to be working perfectly.
Is it possible to replicate data from one publishers to a multiple subscribers in transaction replication? In other words I have Server A, Server B, Server C with databases A,B,C respectively. I need to replicate 10 articles from A to B but only 1 from A to C. In other words I want to use the same publisher but select one article from that publisher instead of all 10 to Server C/database C.
If it is possible what will be the drawbacks of such implementation? Will performance be a hit?
I am Using Transactional Replication with Updatable Subscriptions. Is there any limitations in the number of subscribers (Servers) used for this type of replication?
I have a scenario of configuring this replication in 60 Subscribers(Servers). And the replication should be in the Continuous running mode. Will Transactional Replication with Updatable Subscriptions work in this scenario??? Or is it meant to work for less than 10 subscribers?
Is there a limit on the number of subscribers for Transactional replication?. My question is how many maximum number of subscribers can we have in transactional replication?. How do I retrieve this information throug T-SQL (any DMVs or system tables?). Please let me know.
I am looking to know for both SQL 2005 as well as SQL 2000 and also if anyone can tell me if anything has changed in SQL 2005 for transactional replication that would be great.
We are developing a system which will have to support more than 3000 subscribers. We will have to support both Transactional replication and Merge replication.
I checked the following document about SQL 2005 replication <http:// www.microsoft.com/technet/prodtechnol/sql/2005/mergrepl.mspx>. The document does not clearly specify what is the maximum number of subscribers supported without a significant performance degradation.
The questions i have are: 1. Given the fact that there will be more than 3000 subscribers, there will be more than 500-1000 subscribers trying to replicate at the same time. Will be there be a performance degradtion in such a scenario
2. Has anyone used SQL Server 2005 in a scenario involving more than 3000 subscribers?
3. Will it be better if we develop our own system to perform replication activity instead of relying on SQL Server 2005?
- Ngm
Mail me atnarasimha (DOT) gm (AT) gmail (DOT) com )
I have a problem Setting up pull subscription from a publishser to a subscriber , but I could set up Push subscription with out any problem between these two server .. can any one help me out.
I run SQL Server 2000 Enterprise edition.
I get the following error
"Publication 'ProdDB' does not exist. The step failed."
where ProdDB is the publication database from which my Push subscription is running.
Hi, I have two databases on MS. SQL Server 200 the first one is themain DB and the second one is a replica of it, both of them indifferent territories .I connect them using merge replication justwhen I need to merge the data.The problem is when I make import data or export data from anylocation (MS access ,MS SQL server ) to the second one (subscriber DB)or the main DB(publisher and distributor ) ,the imported data doesnot replicated to the other DB ???Normal transaction replicated smoothly.I hope you can help me.Thanks in advance.Mahmoud Khebaisa.
I have some problem about Merge Replication.i'm trying to merge Database A to Database B in local Server. So Database i want publisher contain this:
Because only Data in Table change therefore so i choosed it 100% Snapshot was generated after that ( Problem not a valid window user i already figured out )After that. I created Local Subscription in same Server ( Pull subscriptions and Subscription type: Client ),now problem is throw out."The schema script 'vwBuyADT_513.sch' could not be propagated to the subscriber."
I tried research so many time in Google but any information i found isn't useful for this problem.This problem still can ignore,Synchronization still running. But after 4-5 Hours running..This messages throw out: "The Publisher failed to allocate a new set of identity ranges for the subscription. This can occur when a Publisher or a republishing Subscriber has run out of identity ranges to allocate to its own Subscribers or when an identity column data type does not support an additional identity range allocation"
My question is: Is there anyway solve 2 probem? : 1. "The schema script 'vwBuyADT_513.sch' could not be propagated to the subscriber." 2. "The Publisher failed to allocate a new set of identity ranges for the subscription. This can occur when a Publisher or a republishing Subscriber has run out of identity ranges to allocate to its own Subscribers or when an identity column data type does not support an additional identity range allocation"
I am using SQL Server 2000 SP4 running on a MS Server 2003 with SP2. I am preforming Merge Replication on a continouas basis and the Merge Agent keeps stopping with the error 203 The process could not enumerate changes at the 'Publisher'. There is no additional information available with this error. Does anyone have any sugestions as to why this is happening. I can manually re-start the agent but it will fail again for particular reason.
I would like to know what alternatives are avaliable to SQL Server Merge Replication. I am also looking for Third Party Tools. Can anyone name a few for me.
Hi, I posted about this before, and set out on my own to get this working, and haven't been able to. I'm trying to get merge replication working with my SQL server 2000, and after 2 weeks I still have nothing. I've gone through multiple 'walkthroughs' which all brought me to the same point. I'm getting down to crunch time, and I'm either going to use this or implement my own merge algorigthm (I'd much rather use this). So here's where I've gotten to:
Hi, I am trying to replicate a production database server,on sql server 2000 at a particular geographic location to a new failover database server on sql server 2000 in a different geographic location via internet. The intention here is to use the failover database during times when the production server is down or busy and the synchronization needs to be sceduled for every 10 min. It will be of huge help if any expert could give the detailed process involved and any precautions that need to be taken. I also need to keep inmind to use the ever/odd sequence number generator during the replication implementation. Also, Whether to use row-level or column-level tracking?
I am having problems with my sql merge replication. Whenever a user syncs up to my main database, most of their records are deleted instead of being merged. Or the records on the main database are inserted and it replaces the whole table with the records on the remote laptops. Is there a way to prevent this from happening? Someone please help me.
I do a merge replication between Sql server 2000 and SQLCE 2.0
On my SQL2000 I have 4 tables i want to merge (specific columns only ) in 1 table for Merge with my SQLCe ( the table will be use for read only)
Question 1: What is the best pratice for keep the information update? Run store procedure before the synch for re-populate the table?:confused: or Make Trigger INSERT, UPDATE, DELETE in the all 4 table?:confused: or a mixte?:confused:
Question 2: Does someone know about some web site talk about this type of trick?
I have installed SQL Server 2005 x64 Enterprise edition with Service Pack 2 on a Windows Server 2003 x64 Standard Edition with Service Pack 2.
Now I have to configure Merge Replication that will work with SQL Server Compact Edition database on Windows Mobile devices.
Distributor and the Publisher are the same server.
IIS 6.0 is installed on the windows server. I have installed the SQL Server Compact Edition Server tools on the server. However the compact edition server tools are only available for 32bit servers and I have also found out from the article http://support.microsoft.com/default.aspx/kb/912430 that you cannot replicate data from SQL Server 2005 to SQL Server Compact Edition by using the 64-bit version of IIS. So if this is true does that mean I can not use merge replication on 64 bit server? Does that mean I have to get another 32 bit server with 32 bit IIS on it to make this work or is there another work around. Am i missing something here?
In a nutshell, I want to do a merge replication with a SQL Server and several Access databases. I haven't been able to find anything in the documentation or 3rd party books.
Hi , I would like to get your advice in an issue that I am facing regarding merge replication. The flow is like: 1) User enter his login name and password. 2) Host_name is set as the login name that the user enters. 3) Based on the HOST_NAME property after filtering user gets the records relevent only to him (Differnet subscriber may use same mobile device).
4) He make some changes in the local sdf according to the business logic. 5) The application Syncs with server on the next login session.
My issue is when I am trying to run the application in emulator exactly at the sdf updating part the application terminates. Error mess: Connection to Remote Device has been lost..... If I execute the same sql insert query it works om the local sdf and master database. Emulator used where Pocket PC 2003 SE and Windows Mobile 6 classic emulator. I have installed 1) Visual Studio 2005 SP 1 2) net compact framework SP2.
When I comment all the code related to Sync ,application works witout any issue. And the same application works in my "I mate kjam".
The code that I use is given below: ==========================================================================
public void Sync(string hostname, int newsub) // Parameters for HOSTNAME and flag to denote new user or existing //user.
I am running SQL Server 2000 and SQL Server CE on the handheld device. I have created the publication on the SQL Server.
I can subscribe to the publication successfully on the handheld and it creates the .sdf file just fine.
When I then attempt to synchronize, the process will run for a long time and eventually error out. (The database has a couple large data tables that are used for value lookups)
The error says the connection was broken.
Any suggestions on correcting this?
I am thinking of breaking the replication database into 2. One with the large lookup files and one with the updated tables. I still need to get the large database with the lookup tables onto the handheld though.
Any thoughts, advice or similar experience handling a situation as this would be appreciated.
I want to transfer data between an sql server 2005 and my pocket pc. So i've installed a sql server CE on it.
I know that i can use the merge replication but, i don't have any IIS. So, i've got to transmit data though email that have got a file attached. I Know how to email from sql server 2005. When the pocket receive that email, i don't know how to insert the data in sql ce from outlook compact. Doi have tu use SSIS , in that case how can i run the package?
I have set up merge replication with 1 publisher and 1 subscriber. Distribution is handled by a 3rd server.
I can generate a snapshot at the publisher and apply it to my subscriber. But when I insert some data (approx 30,000 rows) , the Synchronization agent gives the following error when It runs:
The merge process is retrying a failed operation made to article 'xxx' - Reason: 'The Merge Agent was unable to synchronize the row due to one or more unanticipated errors in the batch of changes. When troubleshooting, increase the -OutputVerboseLevel setting, restart the agent, and check for and resolve any errors generated by the database engine. '.
I have increased the OutputVerboseLevel setting and specified a file path in the -OutputMessageFile but the File is not being populated. All the references on books online say to put the file path in the -Output parameter but when I do it says it can only hold an integer value. So I cant see the errors generated by the database engine.
I'm currently investigating the feasibility of having SQL Mobile on a number of devices running Windows CE. These devices will have to synchronize with the server (either using merge replication or RDA) over GPRS. The problem I have is that some of these devices will never be online. Is there are a way of synching these offline devices? The model I had in mind was using a USB key (or some other storage device) to download the latest updates from an online device and transferring this data to the offline device, and vice versa e.g. from offline to online and then merging. Is this feasible?
"The merge process could not retrieve identity range resource for table..."
then after solving this error we again started syncing of servers manually then we got some other error like
"the process could not enumerate changes to the subscriber".
We tried our best to solve this problem but some error appears each time like
"The process could not deliver the snapshot to the Subscriber."
"The schema script 'sync -t"Design" -o"dbo" -d"\ECOSYSTEMD$FTPMSSQLEcoSystemftpECOSYSTEM_MLEcosystem_MLEcosystem20060619204802Design_1337.bcp" -hORDER( [intDesignId] ASC)' could not be propagated to the subscriber"
again identity erroe emerges and agin we do the same process.
We have a central office with a SQL2005SP4 server (yeah, I know... old as heck) that's the main database and it has multiple subscribers in regional offices. Well... one of the regional offices server is failing, and it needs to be replace.
The original server is an ancient Win2003 86x
The Server team will build a new Win2008r2 64x, and use the same name and IP address
And I'm tasked with the SQL part.
I'll be installing the same version/patch of SQL, but 64x instead, and migrate all databases, including the system databases.
How do I handle replication? Do I need to reintialize from scratch? or can I just use the backup as a starting point?
