Synchronization Comparison between DBTwin and Always On

Synchronization Comparison between DBTwin and Always On

Notes:

3. In DBTwin AA cluster,in order to simplify the maintenance, the

default recovery model is simple.However, it can be modified at

any time if needed.

4. In DBTwin AA cluster,only after the previous DML being executed

succefully on both nodes can the client begin sending the next

query to the database. This is called full sync. On the other

hand,in Always On, the so-called sync mode is actually half sync

because although the log is synced in time,the data is delayed

to sync. So it is called half sync.

6. In DBTwin AA cluster, only DMLs on persistent tables are synced

automatically. if a table is not persistent from the business

perspective, then its DMLs need not be synced. Only a few necessary

DDLs are synced automatically, all other DDLs such as index rebuild,

index organize, auto shrinks, create/update/drop statistics and so

on are all not synced. And what's more, several DDLs such as

create table,create view etc are handled manually. However, in Always

On, all DMLs and DDLs including index rebuild, index organize,

create/update/drop statistics,auto shrinks and many others are all

synced automatically which would bloat the log file and impact the

overall performance.

7. Because of performance impact, all applications with this kind of

behavior or design can NOT run smoothly with Always On in synchronous

mode. For more details:

Case study: A failure of Always On but a success of DBTwin AA Cluster

https://www.zhetides.com/nd.jsp?id=31

8. In Always On due to the trade-off between the sync mode and the

overal performance,the sync mode can be occasionally switched

into async mode automatically. For more details:

Why Always On doesn't have Guaranteed Zero Data Loss(GZDL)?

https://www.zhetides.com/nd.jsp?id=24