White Papers

Overview of Dell EMC SC Series tools for HA/DR
7 SQL Server High Availability and Disaster Recovery with Dell EMC SC Series | CML1119
Note: The term replication in this document refers to SC Series replication. While Microsoft SQL Server does
offer various types of data replication, they are not typically used to satisfy high availability requirements and
therefore not discussed in this document.
1.5.1 Asynchronous replication
Dell EMC SC Series asynchronous replication offers a great amount of flexibility. When the Replicate Active
Snapshot option is not used, replication takes place whenever a snapshot is taken. In this mode, changes are
sent in a batch on a schedule that is defined by the snapshot frequency. When the snapshot is taken
(preferably with Replay Manager) those changes are forwarded to the volumes at the destination.
1.5.2 Semi-synchronous replication
Semi-synchronous replication is a term used to describe asynchronous replication with the Replication Active
Snapshot option. Changes are queued and replicated as soon as they are written on the source. Assuming
the storage network and destination array can keep up, this provides a near zero data loss solution. Semi-
synchronous replication is a very practical solution since it can provide near-real-time synchronization without
the additional latency cost of a synchronous replication solution. Since write order is not guaranteed between
replicated volumes, the recovery point for databases hosted on multiple replicated volumes will likely be the
latest replicated snapshot and recovery based on the active snapshot would be best-effort depending on the
crash recovery mechanism of SQL Server to bring the database to a consistent state.
1.5.3 Synchronous replication
For replication solutions where zero data loss is needed, SC Series synchronous replication may be used.
One of the key differences between synchronous replication and asynchronous replication is that with
synchronous replication, writes to the source storage center must also be replicated to the destination storage
center and acknowledged before the write is acknowledged to the host. This guarantees that the data is
protected. This behavior is provided in the High Consistency mode of synchronous replication. The
guaranteed consistency is provided at the cost of higher overall write latency since the total write latency must
also include the write to and acknowledgment from the secondary array.
High Availability mode allows the source volume to accept writes even if the destination volume is
unavailable. In High Availability mode, if the destination volume is unavailable, replication will essentially
switch to asynchronous replication. When the destination volume is available and resynchronized, it switches
back to synchronous replication. The tradeoff for increased replication resiliency is the potential for data loss if
a catastrophic failure on the primary volume occurs while the secondary volume is not up to date. Since this
potential exists, High Availability mode does not fulfill a zero-data-loss requirement. The paper Dell EMC SC
Series Storage: Synchronous Replication and Live Volume is a great resource to learn more about SC Series
replication.
1.5.4 Understanding replication requirements
When planning a replication architecture, it is important to understand what the costs are to determine the
replication type and the infrastructure required. In some cases, the RPO requirements are set and the
infrastructure is designed to support them. In other cases, the infrastructure may be set and the replication
features that provide the best RPO are chosen. Either way, understanding the data change rate and the
amount of data to replicate is important in determining the design.
SC Series arrays have a replication feature called Simulate Replication that help determine what the actual
cost of replication will be. This feature allows setup and configuration of replication and then tracks and