Saturday, January 15, 2011

HA/Clustered SAN Storage Solution for VMware?

VMware server HA is resolved by "VMware HA" already but what's about HA on storage level?

I mean real storage HA, no single point of failure, switchover without down time, synchronous mirroring between two storages, CDP - continuous data protection.
Have you ever thought about any solution to ensure HA on storage level?
I have spent many hours, days to find out proper solution.

VMware ESX/ESXi hosts can be placed into groups called clusters. A cluster contains 2 to 32 ESX hosts that work cooperatively together to provide features such as VMware HA and DRS. All the virtual machines in this HA cluster are protected by VMware HA. VMware HA detects failures of the ESX hosts, so when an ESX host fails, VMware HA moves and restarts the virtual machines on another unaffected ESX host in the cluster. Shared storage is an important component of the solution. Since the storage array stores the virtual disks of possibly all the virtual machines, even the smallest disruption can create a wide range of effects in a storage array where availability is vital.

I've found solutions: SAN storage cluster or SAN virtualization.
Read more bellow....


SAN Storage cluster:
For both iSCSI and Fibre channel exist clustered SAN storages.

IBM
IBM Storwize v7000 Mid-range SAN Storage - totally virtualized environment
The IBM Storwize V7000 solution is clustered SAN storage consits of controller enclosures each with two controllers and  provides a modular storage system that includes the capability to virtualize external SAN-attached storage as well as its own internal capacity. The IBM Storwize V7000 solution is built upon the IBM SAN Volume Controller technology base and utilizes technology from the IBM System Storage DS8000® family. Storwize V7000 provides an active-active solution and supports Fiber Channel and iSCSI connectivity.

The Storwize V7000 contains at least one controller enclosure. The controller enclosure consists of two node canisters (controllers); the two node canisters are part of a cluster and act as a single processing unit and form an I/O group. The I/O group is the control point for volume access within the Storwize V7000 cluster and each node canister is restricted to a single I/O group as shown bellow.

In addition to the controller enclosure the Storwize V7000 systems may have between one to nine expansion enclosures connected through a redundant serial-attached SCSI (SAS) connections. Each controller enclosure and each expansion enclosure can support a combination of solid-state disk (SSD), SAS 15 K, 10K and nearline SAS drives.

Volume mirroring: a volume can have two physical copies. Each volume copy can belong to a different storage pool, and each copy has the same virtual capacity as the volume.

For more details check IBM Web sites:
http://www-03.ibm.com/systems/storage/disk/storwize_v7000/index.html
http://publib.boulder.ibm.com/infocenter/storwize/ic/topic/com.ibm.storwize.v7000.doc/svc_webgetstartovr_21pax3.html
http://wn.com/Storwize

HP
HP StorageWorks P4500 Multi-Site SAN cluster
HP P4500 Network RAID stripes and mirrors multiple copies of data across a cluster of storage nodes, eliminating any single point of failure in the SAN. Applications have continuous data availability in the event of a power, network, disk, controller, or entire storage node failure.
HP P4500 storage nodes, which are active/active storage nodes, provide highly available shared storage for VMware vSphere environments.
HP P4000 for VMware (Full fault tollerant environment)


For more details check VMware Knowledge Base:
VMware vSphere 4 Update 1 support with HP StorageWorks P4500 Multi-Site cluster

EMC

EMC VPLEX is a new federation solution that can be stretched across 2 geographically dispersed datacenters separated by synchronous distances (maximum distance of separation=100km, maximum round trip latency=5 ms RTT). It provides simultaneous access to storage devices at 2 sites through creation of VPLEX Distributed Virtual Volume. Provided all vMotion over Distance requirements are met, with Distributed Virtual volume configured on VPLEX, a distant VMotion operation can be done from Datacenter 1 and Datacenter 2. You can vMotion over Distance a virtual machine from one datacenter to another to avoid disasters, migrate workloads to save power, or load balance workloads. 
EMC VPLEX for VMware
For more details check VMware Knowledge Base:
vMotion over Distance support with EMC VPLEX Metro

SAN Virtualization Appliances:
Fibre channel virtualization can be used in heterogenous environment to mirror data over more storages and present HA LUN to server.

IBM SAN Virtualization Controller (SVC)

HP SAN Virtualization Services Platform (SVSP)

3 comments:

  1. What solution do you prefer for high available VMware environments?
    For me the winner is IBM Storwize v7000 because provides SAN virtualization, capacity, HA clustered mirrored volumes and performance and can be used on Fiber Channel SAN.

    ReplyDelete
    Replies
    1. On IBM Storwize V7000 you try shutdown a entire i/o group where de caching i/o group of volumes reside and verify whats happens with second copy of volume mirroring ?

      And what is the state of volume presented to hosts ??

      good work

      Best regards

      Delete
  2. On IBM Storwize V7000 you try shutdown a entire i/o group where de caching i/o group of volumes reside and verify whats happens with second copy of volume mirroring ?

    And what is the state of volume presented to hosts ??

    good work

    Best regards

    ReplyDelete