Questions about network RAID10

[flyingJet]

Hello guys.

So I've inherited HPE LeftHand P4000 from previous system admin and I am trying to understand everything about it.

We do have 5 blades with Network Raid 10, fifth one is defined by Virtual IP.  My question is does this fifth blade participate somwhere in the array as hot spare? To my understanding RAID 10 requires even numbers of hhds, for sure I know we are losing one blade worth of space.

Appreciate the answers!

[oikjn]

the storevirtual SAN is grouped as NODES.  You use the CMC program to connect to these nodes and view the "management group" which inside it has one or more clusters which are made up up of one or more of these nodes.  Each cluster can have one or more VIPs (virtual IPs) to connect to for your iSCSI initial connection.

It is not clear about your hardware setup, but if you are using the VSA software nodes, the raw storage redundancy is performed outside of the virtual machines such that any disks the VSA nodes sees are treated as part of a stripe (IE like raid 0).  In the storVritual world this side is typically referred to as the "hardware raid" side.  The "network raid" side, or NR# side is the referring to the node to node redundancy level of a LUN as defined within the cluster group.  Each LUN can be assigned a different Network raid protection level, but NR10 is the default one and should be used for most all situations.  NR10 is similar to hardware raid10, but there are some practical differences since it is really working on the node level and not the disk level.  NR10 means there will be two copies of your data across the cluster.  If you have two nodes that means the data is mirrored on both nodes, but if you have three nodes or any odd number of nodes, the location pattern of the data is not as simple.  That being said, from a planning persepective, there are a few easy guidelines to follow.

1.  the usabable space on each node will be equal to the usable space on the smalled node (IE, three 10TB nodes and one 1TB node in the same cluster would have the same total usable space as three 1TB nodes).

2.  A cluster's max speed is limited by the slowest node in the cluster (IE, placing one new 7200 RPM node in with two other nodes that have SSDs slow down your cluster performance)

3.  When planning on actual usable provisioned space, assume NR10 which means you need 2x the available space in the cluster,.  The easy way to do this when adding a new node is that assume each node you add will effectively increase your usable capacity by 1/2 of its actual capacity.

4.  Use thin provisined LUNs only even if you expect it to become fully utilized as thick LUNs create thick snapshots.