[Phpmyadmin-devel] GSoC Synchronization Project

[Isaac Bennetch]

Hello,

On 3/17/15 3:13 AM, Jay Jayswal wrote:
> Hello,
> i am GSoC Participant. i am interested in Synchronization Project. 
>
> what i understood is we have to build PHP Scripts which connect to
> Remote Machine and Synchronize data.
> which means same database structure at both end and we are coping each
> end's new row's to another end. Right?

Yes
>
> i have started with my proposal. i wanted to ask what should be length
> of proposal? 7 or 8 page is okay or Do i have to make it short?

There's no firm limit on length. I think some of the technical
information for this project in particular will require it to be a bit
longer than others, but having a proposal that's too short (or too long)
is not a disqualifying factor. If we require more information, we can
ask for clarification and you'll be able to revise the proposal until
the submission deadline.
>
> Actually, i already did something like this in my past project.i can't
> disclose code because i signed Non-Disclosure Agreement of that
> company. it was two platform project. web-application(PHP) and Desktop
> application(C# WPF). data is stored at desktop database(SQLite)
> locally, so user can view, update, delete data offline. and when
> internet available, system will run script to synchronize data. 
>
Sounds like useful experience, provided our Non-Disclosure Agreement
doesn't prohibit you from working on this :)

> I have faced many problems while solving the problem that i have
> listed below. can you send me your expected solution for those problem
> so that i can understand your expectations and project requirement.
>
> Problems:
>
> *_Insert_**:*
>
>
> Case 1: insert at DB1
>
> DB1
>
> 	
>
> DB2
>
> Before Synchronization
>
> Step1: Two record with ID 1 & 2 in inserted
>
> Current rows :1,2
>
> 	
>
> Current rows : null
>
> After Synchronization
>
> Current rows :1,2
>
> 	
>
> Two record with ID 1 & 2 in inserted
>
> Current rows :1,2
>
>  
>
>
>
>
>
>
>
>
>
Correct

> Case 2: insert at DB2
>
> DB1
>
> 	
>
> DB2
>
> Before Synchronization
>
> Current rows :1,2
>
> 	
>
> Step1: Two record with ID 3 & 4 in inserted
>
> Current rows :1,2,3,4
>
> After Synchronization
>
> Two record with ID 3 & 4 in inserted
>
> Current rows :1,2,3,4
>
> 	
>
> Current rows :1,2,3,4
>
>
>
>
>
>
>
>
>
>


Correct again

> Case 3: Insert at Both DB1&DB2
>
> DB1
>
> 	
>
> DB2
>
> Before Synchronization
>
> Current rows :1,2,3,4
>
> 	
>
> Current rows :1,2,3,4
>
> Step 1: one new record inserted with id 5
>
> Current rows :1,2,3,4,5(of DB1)
>
> 	
>
> Current rows :1,2,3,4
>
> Current rows :1,2,3,4,5(of DB1)
>
> 	
>
> Step 2: one new record inserted with id 5
>
> Current rows :1,2,3,4,5(of DB2)
>
> After Synchronization
>
>  
>
> 	
>
>  
>
> What should happen in this case? If we want to sync this we need
> some differentiate to identify both DB. 
> In my past project I provided Device ID. Means PRIMARY KEY and DEVICE
> ID both will make Composite Key which make it unique to identify.
>
> What is your preferred solution?
>
>
>
>
>
>
>
>
>
>
>
>
>
>  
>
>
>

This is the difficult solution; my opinion is that in cases like this we
must prompt the user for which side is correct.

>
> *_Update_**:  in this case we required updated row timestamp, I am
> assuming there is update_timestamp column in each table*
>
>
>
> Case 1:  update at DB1
>
> DB1
>
> 	
>
> DB2
>
> Before Synchronization
>
> Step1: updating row with id 1
>
> Current rows :1(updated),2,3,4
>
> 	
>
> Current rows :1(old),2,3,4
>
> After Synchronization
>
> Current rows :1(updated),2,3,4
>
> 	
>
> update row with id 1
>
> Current rows :1(updated),2,3,4
>
>
>
>
>
>
>
>
>
>

Right

> Case 2:  update at DB2
>
> DB1
>
> 	
>
> DB2
>
> Before Synchronization
>
> Current rows :1(updated),2(old),3,4
>
> 	
>
> Step1: updating row with id 2
>
> Current rows :1(updated),2(updated),3,4
>
> After Synchronization
>
> update row with id 2
>
> Current rows :1(updated),2(updated),3,4
>
> 	
>
> Current rows :1(updated),2(updated),3,4
>
>  
>
>
>
>
>
>
>

Right again
>
>
>
> Case 3: update at both side but different timestamp
>
> DB1
>
> 	
>
> DB2
>
> Before Synchronization
>
> Step1: updating row with id 2
>
> Current rows :1,2(updated),3,4
>
> 	
>
> Current rows :1,2(old),3,4
>
> Current rows :1,2(updated),3,4
>
> 	
>
> Step1: updating row with id 2
>
> Current rows :1,2(updated),3,4
>
> After Synchronization (both update is different and timestamp is also
> different)
>
>  
>
> 	
>
>  
>
> In this case both side update is different, what should be final
> record. I think I should be last one updated. Right?
>
> What is preferred solution?
>
>  
>
>
>
>
>
>
>
>
>

In this case, I again think we must prompt the user, but we can also
provide the information that DB2 is more recent.
>
>
>
>
>
> Case 4: update at both side but same timestamp
>
> DB1
>
> 	
>
> DB2
>
> Before Synchronization
>
> Step1: updating row with id 2
>
> Current rows :1,2(updated),3,4
>
> 	
>
> Step1: updating row with id 2
>
> Current rows :1,2(updated),3,4
>
> After Synchronization (both update is different but time stamp is same)
>
>  
>
> 	
>
>  
>
> In this case both side update is different and time stamp is same!! So
> here again we need  DEVICE ID by which we can identify that both
> update is different.
>
> And what should I store finally, because which one we have to finally
> store is depends on timestamp, which is SAME for BOTH DB!!! What is
> preferred solution?
>
>  
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> *_
> _*
>

Probably the same as the Insert case where we prompt the user. Note that
each time we prompt the user, we should also offer them the option to
cancel, synchronize all A to B, and all B to A for other conflicts
during this specific run.

> *_Delete_**:*
>
> Case 1: deleted at DB1
>
> DB1
>
> 	
>
> DB2
>
> Before Synchronization
>
> Step1: row with id 3 is deleted.
>
> Current rows :1,2,4
>
> 	
>
> Current rows :1,2,3,4
>
> After Synchronization
>
> In this case row with id 3 will be added at DB1 because for SYSTEM
> it’s newly added row at DB2.
>
> To handle this problem in my past project I maintained Activity logs.
> Which stores that row with id 3 is deleted, so when synchronization
> take place, I fetch logs from activity log and first delete row with
> ID 3 in DB2.
>
> What is your preferred Solution?
>
> Current rows :1,2,3,4
>
> 	
>
> Current rows :1,2,3,4
>
>  
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
I think the activity log or some form of that it going to be the only
way to do this reliably.

>
>
> Regards,
>
> Jay Jayswal
>

Regards
~isaac
>
>
> ------------------------------------------------------------------------------
> Dive into the World of Parallel Programming The Go Parallel Website, sponsored
> by Intel and developed in partnership with Slashdot Media, is your hub for all
> things parallel software development, from weekly thought leadership blogs to
> news, videos, case studies, tutorials and more. Take a look and join the 
> conversation now. http://goparallel.sourceforge.net/
>
>
> _______________________________________________
> Phpmyadmin-devel mailing list
> Phpmyadmin-devel at lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/phpmyadmin-devel

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.phpmyadmin.net/pipermail/developers/attachments/20150317/6fc3cc08/attachment.html>