White Papers

Preparing SQL Server for SC Series Replication
12 SQL Server High Availability and Disaster Recovery with Dell EMC SC Series | CML1119
3 Preparing SQL Server for SC Series Replication
When using SC Series replication for data protection, in addition to choosing the proper type of replication
there are other considerations that should be taken into account. Paying careful attention to the following
items will assist in performance, efficiency, and recoverability of the replication solution.
3.1.1 Volume and file selection
Since replication occurs at the volume level, all volumes that contain SQL Server data will need to be
replicated. This includes all volumes that contain data files, log files, and any other files that comprise the
database set. Make sure only data that needs to be replicated exists on replicated volumes. If databases exist
on those volumes that do not require replication, they can create unnecessary replication traffic.
3.1.2 Tempdb considerations
Tempdb is used as a temporary work area for SQL Server and is rebuilt, rather than recovered when SQL
Server restarts. This means that while the availability of tempdb is important, data recovery is not. Therefore,
there is no need to replicate tempdb data or log files.
3.1.3 Replay Manager
Replay Manager should be used to take snapshots at regular intervals on replicated volumes. This will enable
reliable recovery from snapshots in addition to the replicated copy of the volume in the event that replication is
out of sync or a prior recovery point is needed. Snapshots taken with Replay Manager are preferred to regular
array-based snapshots since VSS flushes dirty pages from memory and then holds writes while taking the
snapshots, ensuring reliable recovery from snapshots. With array-based snapshots, there is a risk that the
database may not be in a recoverable state.
3.1.4 Database level or instance level protection
It is important to understand that by protecting the database volumes, only selected databases are being
protected, not the entire SQL Server Instance. It is no different than using the SQL Server HA/DR features,
such as Database Mirroring or Always On Availability Groups, in that regard. The message here is that there
are other critical components in the SQL Server Instance that may need to be protected, such as logins or
SQL Agent jobs. A process will be required to accomplish that. While these components are stored in the
system databases that can be backed up, the recovery steps can be somewhat complicated unless the target
SQL Server instance will have identical hardware configuration, instance name, server configuration and
version. Typically, these things may be different at the target site so the most common approach is to protect
these items with deployment or automated scripts.
Another approach to providing instance protection is to replicate all volumes and use the boot from SAN
feature. In this manner the entire server is protected and the replicated volumes could be recovered on an
identical server at the target site.
3.1.5 SQL Server tuning
There are a couple of items that can be considered from a SQL Server perspective to optimize the Dell EMC
SC Series data replication process. The first one is to enable data compression inside the database.
Compressing the SQL Server data pages will reduce overall I/O and therefore reduce the amount of data that
needs to be replicated. Compression benefit and cost will vary depending on the data so while it’s easy to
implement, the performance impact will need to be assessed. Second, evaluate the database applications if
possible to see if there are any efficiencies to be gained. While it can be a difficult task, application and