HP StorageWorks 4000/6000/8000 .HP StorageWorks SAN Design Reference Guide, Pa - Page 294

• Always use the least possible number of copy sets. If possible, combine virtual disks that have the same failover requirements because it is best to have one copy set per application instance. • Copy only data that is more expensive to re-create than to copy. • Add copy sets that will not impact normal data traffic. • Consider adding controller pairs to use available bandwidth effectively. For additional recommendations, see the HP StorageWorks Continuous Access EVA planning guide. Determining the required bandwidth You can determine the required bandwidth for any application. This example explains how to measure the amount of new or changed data. 1. Collect the peak read and write workloads for a given period of time. For Windows operating systems, use a tool such as PERFMON to capture the current performance requirements without Continuous Access EVA operating. Similar tools exist for other operating systems. • At each sample interval, capture reads per second (I/Os per second), read throughput per second (Mb/s), writes per second (I/Os per second), and write throughput per second (Mb/s). • If possible, collect read and write latency data. • Perform the collection by application, capturing the data for each logical unit (device) used by that application. 2. Create a graph of each data set that shows where the peaks occur during the day. • Determine whether the daily average change rate is level or bursty. • Consider how these numbers will increase over the next 12 to 18 months. The results of this scaling process become your design goal. • Determine the values for RPO and RTO. • RPO measures how much data is lost due to a problem at the source site. By definition, an RPO of zero (no data can be lost) requires synchronous replication, regardless of which data replication product you use. • RTO indicates when to start using the recovery site. This measurement includes data about application failover and restart. • For asynchronous Continuous Access EVA, the RPO design space is near zero. • XP Continuous Access Asynchronous supports an RPO from near zero to many hours. • Continuous Access EVA and XP Continuous Access all synchronous replication with an RTO equal to zero. 3. Once the data has been collected: • If the RPO is near zero, use the peak write rate and throughput to estimate the bandwidth you need. For some real-time applications (such as Microsoft Exchange), increase the nl bandwidth between 2 to 10 times this initial estimate due to wait time for link access. • If the RPO is greater than zero, then average the change rate over the RPO interval and use this value as an estimate of the inter-site bandwidth. You might need to increase or decrease this bandwidth, depending on the environment and the amount of time needed to complete the last write of the day before starting the next day's work. NOTE: Because it is difficult to predict data compression before you begin, these calculations do not account for the impact of compression. If you determine that you can compress all data at a constant rate, then use that ratio to reduce the effective throughput required from the link. 294 SAN extension