stretched cluster VM redundancy

[temuri426]

in case of stretched cluster can i have primary and secondary copies only in
Main site, not in DR side?

if the node with primary copy fails , secondary copy on the same site will become primary, at this moment can i trigger creation of new secondary copy to one of the DR nodes?

point is that customer does not need to run VM in DR when primary fails.

[LiamP]

A stretched cluster means a cluster may be located in different physical locations and uses zones to separate.

It does not mean, a Primary and DR site.

So, once zoning is configured, then replicas will be located within each zone.

If a customer had a design of two nodes in one site lets say primary and one node in a DR site i.e. 2 + 1, then both replicas will remain in the primary site. 

[MikeSeden]

As far as VMware is concernered- 

Stretched clusters extend the vSAN cluster from a single data site to two sites for a higher level of availability and intersite load balancing. Stretched clusters are typically deployed in environments where the distance between data centers is limited, such as metropolitan or campus environments.

You can use stretched clusters to manage planned maintenance and avoid disaster scenarios, because maintenance or loss of one site does not affect the overall operation of the cluster. In a stretched cluster configuration, both data sites are active sites. If either site fails, vSAN uses the storage on the other site. vSphere HA restarts any VM that must be restarted on the remaining active site.

A HPE SimpliVity Stretched Cluster-

HPE OmniStack software guarantees that in a healthy and synchronized environment the unexpected loss of one HPE OmniStack System will not lead to data unavailability. Customers have also requested the ability to describe the most likely failure scenario where an entire group of HPE OmniStack System may be lost simultaneously, and for HPE OmniStack software to ensure that customer data remains available.

The HPE Simplivity Stretched Cluster solution works in cooperation with vSphere HA and vSphere DRS to ensure that when one or more HPE OmniStack System failures occur in a physical location, guest virtual machines can be restarted in a second location. This provides for business continuity with failover time measured in seconds.

So to speak to: "point is that customer does not need to run VM in DR when primary fails." DR is usually implemented when the possibility of a site failure might happen.  If the Primary site has the Primary and Secondary node and only one node goes down, DR would not be triggered because the other node would do as it is designed and trigger a new VM. Both primary and secondary copies are at the Primary site as described in Liam's 2+1 comment. If the SITE fails then DR would be triggered and copies of VMs would start to come up at the DR site.

So do you need to run DR with SimpliVity? Only if there uis a chance of Site Failure.

[temuri426]

Hi,

yes you are right, customer needs stretched cluster but they to run DR with SimpliVity Only if there is a chance of Site Failure.

is this possible? if yes can please give me bit more technical details, or any documentation which describes the VM copy process with stretched clusters

[MikeSeden]

How many nodes in the cluster?

Look Here- http://www.vhersey.com/2018/02/19/hpe-simplivity-federation-design/

[temuri426]

we will have four nodes in main site and four nodes in DR.

[MikeSeden]

As Liam mentioned you can set up a 4+4 cluster that is zoned so that you have a primary site and a HA DR site using VMware HA and DRS in conjunction with SimpliVity IWO.  The Primary and secondary would each be on one of the 4 nodes for the VM and a HA type copy would be sent to on of the 4 Nodes in the DR cluster. We also have something called SimpliVity Rapid DR which would allow you to choose more specifically with SimpliVity.

https://h20195.www2.hpe.com/v2/getdocument.aspx?docname=a00018755enw 

and here

https://support.hpe.com/hpsc/doc/public/display?docId=emr_na-sv760_000180_aen_us&docLocale=en_US

[temuri426]

Hello,

i as i understand if i create two zones e.g primary zone and DR zone, it means that one specific VM will have its replica in primary zone and secondary zone also ? 

can you please provide an documentation for zonning? need to understand it more deeply

[DeclanOR]

Hi @temuri426 

The easiest way to picture a stretch cluster in my opinion is as follows:

Lets say that you have two zones: Zone A, and Zone B

These two zones can reside in the same datacenter or even potentially in two different physical locations/buildings. (assuming network configuration and latency values allow)

Any VM's that have their primary copy on Zone A, will have their secondary data copy located in zone B, and any primary copies in Zone B will have their secondary data copies in zone A. In the event of a ctastrophic event which brings the entire site A down, then your site B will keep owenrship of it's original primary data copies, but also now take ownership of the secondary data copies of any VM's which were running on site A, ensuring business continuity.

The placement of your arbiter VM will be critical also. The arbiter should reside outside of the hosts which make up your stretched cluster....i.e : do not run your arbiter VM on any host which is part of the stretch.

Hope this helps.

DeclanOR

[temuri426]

thanks for such detail explanation, one more question:

can i have zone A and zone B in main datacenter and zone C in DR?
purpose of this is , if node in zone A fails, VMs should be migrated to zone B in another node but same location, in case of disaster VMs should be migrated in zone C in DR. can i have this kind of design?

if node fail