How Thin Image works

This topic explains how Thin Image works.

Parent Topic

• Overview of Thin Image

  With Thin Image (HTI), you can perform cost-effective replication by storing the differential data between primary volumes (P-VOLs) and secondary volumes (S-VOLs) of virtual volumes (V-VOLs). You can also copy data of an entire volume, instead of copying just the differential data, to a volume. By copying the entire volume, you can expect a higher performance than by just storing the differential data.

Child Topics

• How Thin Image uses pools and pool volumes

  Thin Image stores snapshot data in DP pools (Pool Type: Dynamic Provisioning) or Thin Image pools (Pool Type: Thin Image). Unless they are defined otherwise, both DP pools and Thin Image pools are referred to as "pools". A pool consists of multiple pool volumes (pool-VOLs) which are, as a group, the container for the snapshot data.

• Usage level rebalancing among parity groups

  If multiple parity groups that contain LDEVs used as pool-VOLs exist, rebalancing can improve biased usage rates in parity groups.

• How Thin Image uses V-VOLs

  V-VOLs are virtual volumes that do not have any physical storage space.

• How Thin Image pairs are created

• Using snapshot pairs (not cascaded)

  The following figure shows how snapshot pairs are used when they are not cascaded.

• Using cascaded pairs
• Workflow for creating groups and storing snapshot data using CCI

  You can use CCI raidcom commands to create a consistency or snapshot group and to split pairs to store the snapshot data for the group.

• Methods of storing snapshot data

  The Copy-After-Write (CAW) method is the default method of storing snapshot data. However, if the write pending rate (percentage of data waiting to be written in cache memory) is 60% or higher when writing in cascaded pairs random access mode, or 30% or higher when writing in sequential access mode, the Copy-On-Write (COW) method is used.

• Thin Image pair restoration

  You can use Thin Image to overwrite snapshot data in pools to P-VOLs so that the P-VOL content is returned to the condition it was when you stored snapshot data.