This details the testing environment and the configuration that was employed.

We used the following environment for testing Storwize HyperSwap with IBM i:

We used IBM i level 7.3 Technology Refresh 3

For the tests we used NPIV connection of Storwize to IBM i

We used two VIOS, both of them level 2.2.6.10

Four Brocade Fibre Channel switches are used. They are divided into a public and private switch (via virtual switch/fabric) to provide a private fabric for internal communication between the nodes and a public SAN for all other I/O traffic.

We used two fabrics (A and B) for public and private SAN.

For the tests with PowerHA for i and IASP we used one fabric to make setup for these tests faster and simpler. However we certainly recommend to use two fabrics also for solutions with PowerHA for i.

The two Storwize V7000 storage servers are set up in a HyperSwap topology.

Figure 2-1 on page 7 shows the Server, Storage and SAN connection scheme.

The IBM Storwize V7000 HyperSwap setup consists of two controller enclosures with one expansion enclosure each. The controller enclosures are connected only to the switches in their location or site, there is no cross cabling to the other site. Due to the fact that only four ports are available per single controller each port is connected to a different switch:

This can be extended if more ports are available.

The Storwize V7000 control enclosures build a single cluster, so they have to be clustered in advance. This results in a two I/O group setup, one I/O group per site, which is the minimum setup. At the third site, an IBM DS8000 Storage system is used as the quorum, which has to be attached to both sites, independent of each other.