Table of Contents
Advertisement
Config Guide
Note: If you try to disable overcommit and the total space allocated to thin-provisioned volumes exceeds the physical
capacity of their storage pool, an error will say that there is insufficient free disk space to complete the operation and
overcommit will remain enabled.
Thresholds and Notifications
Users with a manage role can view and change settings that affect the thresholds and corresponding notifications for
each storage pool.
Low Threshold – When this percentage of pool capacity has been used, Informational event 462 is generated to notify
the administrator. Thus value must be less than the Mid Threshold value. The default is 25%.
Mid Threshold – When this percentage of pool capacity has been used, event 462 is generated to notify the
administrator to add capacity to the pool. This value must be between the Low Threshold and High Threshold values.
The default is 50%. If the commitment setting is enabled, the event has Informational severity; if the over-
commitment setting is disabled, the event has Warning severity.
High Threshold – When this percentage of pool capacity has been used, Critical event 462 is generated to alert the
administrator that it is critical to add capacity to the pool. This value is automatically calculated based on the available
capacity of the pool minus reserved space.
8.5. Performance Volumes/LUNs and Rapid Rebuilds:
8.5.1.
Large LUNs:
The RealStor firmware release introduces the concept of Virtual Disk Pools – which are comprised of multiple Virtual
Disk Groups (a disk group is similar to a VDisk). This now decouples the LUN from being restricted to a single
VDisk. A Volume's data on a given LUN can now expand across all disk drives in a pool. When capacity is added to
a system, the user is also getting a performance benefit of the additional spindles –hence the term Performance
Volume.
General "Best Practice" for VMFS datastores is to use large LUNs for use by multiple VMs. Large LUNs gives
VMware vSphere users the most flexibility by not requiring storage administrators to provision new storage every time
a new VM is created. Large LUNs also minimize rescans for new LUNs on the ESX server, which can impact VMFS
I/O.
Operating Systems now support up to 16TB file systems.
As enterprise workloads for virtual environments grow in size and in performance demands, virtual hard disk (VHD)
formats need to accommodate them. Hyper-V in Windows Server 2012 introduces a new version of the VHD format
called VHDX which supports virtual disks up to 64TB in size.
RealStor supports large, flexible Volumes with sizes up to 128TB spanning up to 96 HDDs (architectural limit)
providing seamless capacity expansion. Data automatically reflows to new drives and IOPs scale with additional
storage.
8.5.2.
RealQuick: Rapid Rebuilds
Rebuild of conventional (Linear) storage requires that the entire VDisk is rebuilt prior to returning the VDisk (and thus
the Volumes on the VDisk) to a fault tolerant state. The rebuild operates at the VDisk level and has no knowledge of
Volumes or customer data contained therein.
A customer could have a single small Volume on a large VDisk. During the rebuild, even after it has completed the
stripes where the Volume resides, the VDisk remains in a degraded or critical state. During this period, another drive
failure would take the VDisk from degraded to critical, or from critical to down. In the latter case, the customer may
have lost data. In the prior case, the customer's data becomes at risk once again, even though that area of the VDisk
had been rebuilt.
In another example, the customer may have created a Volume or Volumes consuming the entire VDisk space, but may
have actually allocated very little. Since linear storage has no means to know what stripes contain customer data and
83-00006911-00-01 Rev B
AssuredSAN 3xx4
Page 23
Advertisement
Table of Contents
