Basic Calculations - HP 12000 Design Manual

The most likely scenario for enterprise data centers using HP VLS replication is that it will be
enabled to run 24 hours a day on dedicated links used only for replication. HP VLS has the
ability to enable windows for replication to further customize the implementation to best suit
your needs.
With HP Accelerated deduplication there is a time delay after the backup completes for the
deduplication to run and the replication instructions to be generated before new data can
replicate. This means there is probably a stronger need for a dedicated link and the full
utilization of that link in order to get the backup deduplication and replication completed
within an acceptable service level agreement.
HP also provides at Backup Sizing Tool to help you size any backup solution including deduplication
and replication. See http://www.hp.com/go/storageworks/sizer. Wherever possible, do a proof
of concept before implementing into production.

Basic Calculations

The time required for replication is derived from the equation:
NOTE: You should always factor in additional bandwidth to the calculated link size to take
possible link downtime or replication initialization delays into account. If the replication falls behind
for any reason (for example, the link is down for 24 hours), you need more than the normal daily
bandwidth to allow replication to "catch up." Also, when new server clients are added to the
backup system, their first backup goes over the link wholesale thus causing a temporary increase
in daily bandwidth. HP recommends sizing the link with an additional 25% extra bandwidth to
handle these cases.
The calculations below are based on several assumptions and represent worst case scenarios.
A pessimistic compression ratio of 1.5:1.
Links are only 66% efficient (1 Gb/sec link can typically deliver 80 MB/sec).
Because the VLS is a multi-node architecture with significant performance for backup,
deduplication, and replication, the VLS will never be the bottleneck with respect to getting
data on the wire for replication.
Target usage model is backup and replication complete within 24 hours.
Allow a four hour delay after backup completes for the first set of replication instructions to
be produced.
There is a "typical" split of 75% database, 25% files system data. File system incrementals
are used which will result in larger amounts of data going across the replication links. Database
fulls are used.
Each example illustrates a unique point:
1. Sizing Scenario: 20 TB of database and file-system replication.
1 16 Replication