Had a case opened with Microsoft regarding to a bug in Sync Framework 2.1, similar to the one discussed here:
Requested a hotfix or an update on future development plans. Got this response:
Yes we confirmed that this problem is in SyncFx 2.1 and there are no plans to address it in the framework as future development will be halted for that technology. We plan to address the issue in SQL Azure Data Sync CTP3 CU5, which is the next release.
HUH??? Does this mean that the Sync Framework is being discontinued, and only the Azure Data Sync Agent will remain, without the actual API usable by the developers? Anyone who's familiar with the Microsoft roadmap in this area, please clarify the statement.
Essentially, we are looking to have a reliable framework which can be used to develop custom sync services which sync data from multiple client databases with the Azure "mother" database. Is it still tenable proposition within the scope of the related Microsoft technologies, or we got to consider writing our own sync mechanism?
We are in process of addressing the negative timestamp value issue and will release a hotfix of Sync Framework 2.1 for this issue. We remain committed to make sure that Sync Framework will serve the needs of developers writing applications based on SyncFx. We are now in the planning process of putting together a 18-month roadmap for Sync Framework. If you have specific feature requests, we would love to hear from you.
For your specific scenario of syncing from multiple on-prem SQL Server databases to the hub SQL Azure database, we believe Data Sync Service should be well-positioned to serve the basic needs without writing code and minimal time to deploy; however, if there are complex business logic involved (e.g. custom conflict resolution policies) that cannot be handled by Data Sync Service, then Sync Framework is still the recommended (albeit more costly) approach.
I hope this addresses your concern.
Shirley, thank you very much for your prompt response. It sounds encouraging, not like the disappointing and somewhat patronizing response we received from the support people (see attached below).
Here's a bit more information about our vision, if you like. Perhaps, using the Data Sync Service is not the best idea for what we are trying to achieve. Our scenario involves numerous client applications (installed at different customer sites) which share a portion of their total data via a cloud-based master database. The client databases are SQL Express 2005 or SQL 2008. For the security and other reasons, we envisioned having a light-weight pre-configured custom sync service (built using SyncFX) installed at each client location, which would exchange data with the the master database via a secure WCF-based API. We are talking about thousands of clients in this scenario. Please let me know if you see any problem with the offered vision and if you think that SyncFX + Azure may not be the best choice for the envisioned architecture.
At this moment we already have a live prototype. While deploying it at a test site we encountered the issue of the negative timestamp. We found an expensive and cumbersome workaround, which, however, is not tenable in most of the typical configurations. What we have on the discussion table now is a multi-million dollar decision about whether we are going ahead with the SynFX + Azure, or we should look for something else. Having the response from your support which implied that the development team might not be willing to fix some obvious and a critical bugs - for the reason of cost or whatever, has put a big question mark on the SyncFX+Azure approach.
However, as we've already made significant investment in this area, we would be very thankful for any real support that Microsoft could offer us at this point, and primarily with the assurance that this particular bug will be fixed and soon, and there will be a full commitment to supporting the technology - meaning at least fixing the critical issues in what's available now or offering viable alternatives in the future. I hope this our concern is also of a concern to the SyncFX/Azure team.
I also second the Mr. Davis' request - it is very important to know WHEN the update is going to be available. Releasing it rather sooner than later will confirm your team's commitment to the technology and supporting us - Microsoft partners (who actually pay significant money to Microsoft every year in order to have such support).
PS. Attached below is what we received from the support team's email yesterday. Besides, our original request did mention the Azure SQL Data Sync Client only because we found that it was producing exactly the same error as our custom-built sync service. Using the Azure Sync Client in our scenario has never been the intent. Therefore, fixing the issue in the Azure Sync Client and not fixing it in the underlying SyncFX is not a solution from our perspective.
I will try my best to explain here. Any potential fix is at the discretion of the product group. Some fixes they OK and some they do not for various reasons. Like how complicated the workaround would be, monetary losses, etc. I think in this case since your original inquiry was regarding SQL Azure Data Sync (this is also where our energies are focused for the future) and you have a workaround, that a fix in Data Sync would suffice.
So that being said, we are supporting 2.1. Because this fix was rejected does not mean that all will have the same result.
this is not the first time that this particular bug has been raised and a question on whether the fix applied to the Data Sync database provider should be applied to Sync Fx as well.
isn't opening a support case with MS "a good business justification" to get the hotfix backported to Sync Fx, especially when this has been a long confirmed bug?
Now, that SQL Azure Data Sync Service supports scope updates, is the team considering back porting that to Sync Fx as well?
The Sync Fx feature voting site used to be with the SQL Data Sync Service feature voting site, it was cleared and only the Data Sync Service feature voting site is up. it will be good if you could have a new feature voting site exclusively for Sync Fx where people can submit feature requests that you can take into consideration for your plans for the product.
Update. This hot fix has been released. The following support page provides detailed information and hotfix download.
Microsoft stays committed to serve Sync Framework application developers' needs. We're currently planning an incremental release in the next 6-9 month timeframe. If you have any specific feature requests, we're here to listen.
- Proposed as answer by Cathy He [MSFT] Friday, June 15, 2012 11:19 PM