VAAI and why Storage Admins should know about it (Part 2 of 2)

Hi Folks. 

Finally - I have been able to upload part 2 of 2 of the series. Sorry for the delay (and thanks to those who sent emails requesting it. Means people are reading). 

THe second part of the series concentrates on BLock Zeroing and Hardware assisted locking. Enjoy -

Raj (rajs@ca.ibm.com)

--------------------------------------------------

Block Zeroing

Block zeroing is designed reduce the amount of processor and storage IO required to write zeroes across an entire Eager Zeroed Thick (EZT) VMDK when it is created.  With the block zeroing primitive, zeroing operation for EZT VMDK files are offloaded to the XiV storage system without the host having to issue several commands

Benefit: Reduce the amount of CPU effort and IOPs required to write zeroes across an entire Eager Zeroed Thick (EZT) VMDK

Hardware Assisted Locking

It  is important to understand why locking occurs in the first place.  For block storage environments, VMware datastores are formatted with the VMFS filesystem.  VMFS is a clustered filesystem which uses SCSI reservations to handle distributive lock management Although normal IO operations do not require this mechanism, these boundary conditions have become more common as features such as Vmotion with DRS are used more frequently.  With hardware assisted locking, the LUN locking processing is transferred to the storage system.  This reduces the number of commands required to access a lock, provides locks to be more granular than the entire LUN, and leads to better scalability of the virtual infrastructure. 

Benefit: Hardware assisted locking will increase VMs per datastore, ESX servers per datastore, and overall performance

IBM Support for VAAI

VAAI is a key enabler to truly gaining the operational benefits of VMware.  Although offered by more and more storage vendors, different storage hardware architectures will lend themselves far better to VAAI integration than others (mileage will vary ).  Enterprise array architectures will leverage all of the  VAAI capabilities while lower level arrays may not be able to handle the additional tasks assigned to them by VAAI. This is why many vendors are only offering support for parts of the VAAI API’s.

IBM is offering VAAI support on  a variety of its disk platforms. The IBM XIV, Storwize v7000, and San Volume Controller (SVC), will support all 3 of the VAAI primitives. Each enterprise class system has specific features that can best leverage VAAI. Distributed Distribution Grid (XIV) , pre-fetch algorithms (XIV/v700/SVC), , storage pooling for more disk spindles (XIV/v700/SVC), load balancing (XIV/v700/SVC), thin provisioning (XIV/v700/SVC), and other high-end features  have made them excellent for VAAI functionality.

IBM has been  working with the VMware Reference Architecture Lab for many years and will continue to partner with VMWare to create the next set of API’s. Future integration  like the VMware vStorage APIs for Storage Awareness (VASA) are expected to improve visibility into the physical storage as well as the virtual data store through vCenter and Capacity IQ.  IBM Storage continues to be an excellent choice for VMware environments with support for VAAI, vCentre Storage Provisioning Plugin’s, SRM, Storage vMotion and future VMware storage enhancements.