White Papers

Data Progression and Hyper-V
38 Dell EMC SC Series: Microsoft Hyper-V Best Practices | CML1009
4.1.1 Data Progression with archival data
In this example, a Hyper-V VM workload creates a large amount of archival data, such as image files or video.
This data is stored on separate virtual hard disks on one or more dedicated CSVs, and once this data has
been written to disk, it is infrequently accessed.
• Option 1: Leverage the Recommended (All Tiers) storage profile. New writes will go to tier 1, and
over time (about 12 days between each tier) Data Progression will move the data to tier 2 (if a tier 2
exits) and then to tier 3. If writing a large amount of archival data to tier 1 does not negatively impact
the performance or capacity of tier 1 that might be needed for other workloads, then the
recommended storage profile would work well.
• Option 2: Configure the CSV to use only the Low Priority (Tier 3) storage profile. This will ensure
that all new data to the CSV is written to tier 3 from the start. This is helpful when the performance of
tier 1 is needed for other workloads or has a limited capacity that would be negatively impacted by
ingesting a lot of new data that is essentially archival once written. With this design, tier 3 would need
to ingest data while maintaining adequate application performance.
• Option 3: Create a custom storage profile that includes tier 2 and tier 3 only. This assumes the array
has three tiers of storage. If the performance of tier 3 is inadequate for ingesting the new data, this
ensures that tier 2 receives the data. The data is kept out of tier 1, and Data Progression will
eventually move the data to from tier 2 to tier 3 (about 12 days).
Create a custom storage profile
This strategy can also be applied to workload elements that require the maximum performance of a higher
tier. Dedicate a CSV to the virtual hard disks that host these elements and select the High Priority (tier 1)
storage profile to keep the data in tier 1 or create a custom profile that allows tier 1 or tier 2, but not tier 3.
This might also apply to volumes containing gold images from which many VMs or hosts will be provisioned.
4.1.2 Data copies and migrations
When copying or moving data from one location to another, it is possible to inadvertently consume all the
available capacity in tier 1. This is because by default, new data is written using RAID 10 in the highest tier,
and tier 1 is often comprised of a smaller number of lower-capacity high-performance disks.
Filling up tier 1 is undesirable because new writes are then forced to occur in a lower tier which can result in
significantly degraded performance for the SC Series storage and any hosted workloads. Normally, alert
thresholds would notify administrators with enough lead time to remedy a tier 1 capacity issue (by adding
disks for example), but a large data copy or migration operation might consume tier 1 capacity before an
administrator has time to respond.