White Papers

Best practices for Live Migrate and Volume Advisor
11 Dell EMC SC Series Live Migrate and Volume Advisor | 3186-BP-SC
4 Best practices for Live Migrate and Volume Advisor
Live Migrate is a simple-to-use feature of Dell Storage Manager that makes it easier to efficiently use and
manage storage. However, it is important to understand what is occurring in the background, as discussed in
section 1.1, Live Migrate states. Use the following best practices to help ensure smooth and efficient
movement of data.
Volume selection: While Live Migrate makes it easier to move a data workload from one SC Series array to
another, this does not mean that workloads should be moved without consideration. After all, it still takes time
and resources to move the data. Planning and attention should be exercised when selecting which workload
to migrate. When choosing between several volumes to migrate, select the volume with the least amount of
storage in tier 1, RAID 5. When Live Migrate replicates a volume, it leverages Live Volume asynchronous
replication to perform the data movement, which by default replicates the data to the lowest tier available on
the destination array. Depending on its configuration, the read performance of the array could be lower
because reads would come from tier 3. Writes on the destination SC Series array will go to tier 1, RAID 10 as
normal.
Note: Enabling the option Edit Data Collector Settings > Replication Settings > Allow Select to Lowest
Tier on Replication Create provides the option to Replicate Storage to Lowest Tier when configuring Live
Migrate. If this default option is not selected when creating a Live Migrate, the replicated volume on the
destination array will be placed into tier 1, and Data Progression will move inactive or less-active pages to
lower tiers over time. Care should be taken in selecting this option because it may cause tier 1 to fill up, which
would result in some writes being sent to lower-performing tiers of storage.
Monitoring performance: Because Live Migrate is not a part of the typical day-to-day I/O activity, it is
important to monitor the impact this additional workload has on the environment and ensure it does not impact
application performance. If either SC Series array becomes overburdened, corrective action should be taken.
In most circumstances, adjusting the source replication QoS node to a lower percentage is sufficient
corrective action, although this may take longer. In some circumstances, it may be necessary to perform Live
Migrate during a period of lower I/O.
Time of day: Live Migrate is nondisruptive with regards to interruption of I/O or connectivity. However, the
data replication, particularly the replication of current writes, places an additional load on both arrays and the
switch fabric. Choosing to perform Live Migrate at a time of lower workload (for example, outside of normal
business hours) will lower the impact this additional workload may have on other applications. It is also
important to consider other concurrent replications and replication QoS node settings.
Transport type: Replication between SC Series arrays can occur over any transport type they have in
common. Local replication can therefore utilize iSCSI, FC, and FCoE protocols to transport the data. This is
unlike remote replication which typically will only have iSCSI. It is recommended that administrators explicitly
select a transport type, and not select All Available Transports. This ensures that replication will occur over
a known route, and greatly eases troubleshooting should issues occur.
Replication performance: Regardless of the selected transport type, steps should be taken to ensure that
sufficient bandwidth is available on the switch fabric (Ethernet or Fibre Channel), and that Dell EMC and
industry best practices have been followed in creating the switch fabric. If utilizing iSCSI for replication,
ensure that both arrays are configured with the same packet frame size and that the switching fabric meets or
exceeds this.
Replication QoS: Replication QoS enables administrators to limit the amount of bandwidth consumed at a
particular time of day or day of the week. All replications occurring at a point in time that are configured to use