Data Domain Sizing

Data Domain sizing is done using the EBSS, and as per usual - always download the latest version.

The steps for sizing a DD are:

1. Pre-sizing preparation
2. Size for capacity
3. Verify and adjust for ingest performance requirements
4. Address any replication needs

Some data needs to be collected to answer the following questions:

1. How much is a full backup?
2. What does the customer's backup software report for a single backup of Exchange, File Systems, databases, etc?
3. How much data is going offsite? - this addresses the retention and replication issues

To select an appropriate model of DD, determine the type, size, schedule and compression rate of the data (or, How much?) and at the same time determine how long data needs to be saved.  The longer the retention period, the greater compression ratio will be realized because you find more commonality.  Increasing the retention from 2 weeks to 1 month to 2-3 months will have a 5x-10x-20x impact on compression.

Burn rate is another factor to consider.  This is the amount of disk actually being used to store the data over time.  For instance, if you have a 1TB backup that, after compression consumes 200GB physical disk, and based on daily change rate we determine that subsequent incremental backups will be about 100 GB and consume 20 GB on disk, our burn rate after a week would be 200GB + (7*20) for a total of 340GB.  This info can also be gathered form the DD system once it is in place as "Post-Comp (GiB)" reading in the backup reports.

To compute the throughput requirements, divide the largest backup by the backup time window. 

Best practice is to keep burn rate and throughput at 75-85% to allow sufficient headroom and growth capability.

Once the data is collected and analyzed, select a DD model by determining what model gives the appropriate throughput in TB/hour. 

You can also calculate by the performance buffer by dividing the max throughput by max capacity and multiplying by 100 to get the % Max Throughput.  Then put that number in the 75-85% buffer to determine your DD.

Remember that the sizing results from the EBSS are the minimum requirement.  Plan accordingly for growth.  Also, performance results are based on the slowest transfer protocol selected.  The EBSS will report that the 1st week compression rates will be minimal (2-4x) due to a lack of commonality, but that they will increase afterward.  This is accurate with the real world.  EBSS also uses pre-programmed compression ration based on a 2 week retention minimum where there will be little or no compression on the first week and after that first week there will be an increase.  Remember this when discussing and reviewing results with a new customer.