PCNS 3.1 for Virtualisation - Solution for multiple Site Config

[BillP]

Hello,

one of my customers have following Environment:

- 2 separate Buildings / Sites (Site A und Site B)

- on each Site are 4 ESXi 5.5 Hosts

- All 8 ESXi 5.5 Hosts are in a VMware HA Cluster

- on each Site is a APC Smart UPS 15kVA

- on Site B is a FibreChannel Storage for the ESXi Hosts

- the UPS on Site A has 8 Battery-Packs

- the UPS on Site B has 16 Battery-Packs (longer Battery runtime)

My Goal is to achive the best Solution for my customer. Of course.

We want to achive that for following Scenarios:

- Site A loosing Power

- Site B loosing Power

- Site A + B loosing Power at the same Time

- Site A + B loosing Power at different Time

I want to use the UPS Configuration "Advanced (VM Migration Support)". I have a physical Server on each Site with Windows Server 2008 R2. This Server and the ESXi Hosts on each Site are powered by the same UPS.

I installed on the physical Server on Site B "PCNS + VMware Support". On the PCNS i added the ESXi Hosts to the UPS at there Site. I configured the Migration and Shutdown Delays for my Environment. After that i configured the Event "Runtime remaining below threshold" for 1500 Seconds. Under "UPS Configuration" i changed the Settings for one UPS. UPS on Site A should do a VM Migration befor shutting down anything. UPS on Site B should not migrate to Site A because on Site B is my VMware Storage. The VMware Environment can't live without the storage. So i want to shutdown the VMs.

If i unterstand right, this Installation has some "Problems".

- If Site B is on Battery, the physical Server just shutdown the ESXi Hosts on Site B. Because the Storage is on Site B to i want to shutdown the VMs (and ESXi Hosts) on Site A, too.

- If Site A is on Battery, the physical Server migrates the VMs to Site B and shutdown the ESXi Hosts on Site A. After that, the physical Server is shutting down himself. In case Site B is on Battery few Minutes after Site A goes down, no PCNS Server could shutdown my VMs and Hosts on Site B.

You can see there are some "Problems" i want to solve. Here i Need your help, please.

Any suggestions, tips or anything else?

Many thanks for every Post.

Best regards

Marco

[BillP]

Hi,

Application Note 180 discusses advanced set up and configurations.

[BillP]

Hi Bill,

thank you for your Reply.

Unfortunatelly the Document is not really clear for me.

No Scenario in the document fits to my situation for 100%.

Under the "Setup 2" is written each VMware Host is protected by a seperate UPS.

In my case a pair of 4 VMware Hosts is protected by a separate UPS. The physical Server with installed PCNS + VMware Support do not have a separate UPS like shown on page 6. Additionally the Storage has his own UPS, too.

So the Setup is not fits not 1:1.

Possible Solution:

Buy a seperate UPS (with NMC) for my physical Server with installed PCNS + VMware Support. So i achieve, the server whould not shutdown if just one of the Site A or Site B UPS on battery and the server can monitor the other remaining Site. To ensure all VMs of alle Hosts will shutdown if a "On Battery" Event occurs in Site B (with the Storage), i will install an PCNS 3.1 Virtual Appliance on Site A with "Single UPS Configuration".

In this Configuration (i believe) Site A is migrated to Site B if Site A is on battery. The physical PCNS Server is still online. All VMs now running on Site B. If now Site B is going on battery the physical PCNS Server will shutdown all VMs on Site B. That is one goal to achieve.

Another Scenario whould be if Site B is going on battery. VMs and Hosts on Site B shuts down (because i disabled Migration for the UPS on Site B). The deployed PCNS VA which is running on Site A will receive a shutdown Signal from the UPS on Site B and will shutdown the VMs and Hosts on Site A, too.

Is this maybe a possible Solution?

Do i unterstand the fact how PCNS 3.1 with VMware works right?

Thanks!

Best Regards

Marco

[BillP]

Hi Marco,

1. I would suggest getting a separate UPS for the SAN in Site B.

2. Install PCNS on physical machine in Site B with VMware Support enabled.

3. Configure PCNS for Advanced Configuration in Site B.

4. Register with the UPS in Site A and the 2 UPS's in Site B.

5. Associate the ESXi hosts in Site A to the UPS in Site A.

6. Associate the ESXi hosts in Site B with their UPS.

7. Do not associate any ESXi hosts (VMware Host Protection page) with the UPS that is powering

the SAN.

8. Enable the option to Shutdown VMware Hosts and PowerChute machine for the UPS powering the

SAN.

9. Disable VM Migration for the UPSs powering the ESXi hosts in site B.

10. Enable VM migration for the UPS powering the ESXi hosts in Site A.

11. Install PCNS on the physical server in Site A (if that server needs UPS protection also) with VMware Support enabled.

12. Configure that instance of PCNS for Advanced UPS Configuration and register with the UPS in site A and the 2 UPS's in Site B.

13. Associate the ESXi hosts in Site A with the UPS in Site A.

14. Associate the ESXi hosts in Site B with the UPS in Site B.

15. Do not associate any ESXi hosts (VMware Host Protection page) with the UPS that is powering

the SAN.

16. Enable the option to Shutdown VMware Hosts and PowerChute machine for the UPS powering the

SAN.

15. Enable VM migration for the UPS powering the ESXi hosts in Site A only.

Configure VM shutdown delays and shutdown event actions on both instances of PCNS as required.

If Power is lost in Site A PCNS will migrate the VMs to ESXi hosts in Site B and shut down the ESXi hosts in site A followed by the physical machine in Site A. The ESXi

host in site B and their VMs are still protected.

If Power is lost in Site B PCNS will shutdown the VMs in Site A and in Site B, the ESXi hosts in Site A and B and the physical machines in site A and B.

If Power is lost in site A and B at the same time PCNS first issues a command to place the hosts into maintenance mode - this will prevent any attempt to migrate the VMs

from Site A to Site B. PCNS will then start shutting down VMs followed by the ESXi hosts in site A and B. It will then shut down the physical machines in Site A and B.

I believe that the above approach meets most of the requirements. When power goes in site B the ESXi hosts in Site A will also get shut down (unnecessarily since the UPS in Site A is unaffected) but then there are no VMs running since the SAN is unavailable. There should be no issue if power goes in both sites simultaneously with both instances of PCNS attempting to shut down the same VMs/ESXi hosts.

With the current design of PCNS 3.1 the scenarios supported in the Application Note do not match your customer's environment. However this scenario is something we will try to cater for in a future release of PCNS.

[BillP]

Hello,

thank you for your reply.

I tried your Setup last week. But with no success.

I have two APC Smart UPS 15kVA on both sites. I got an third UPS for my Storage (APC Smart UPS 3000VA Rack). If i try to add all three UPS on my physical Server on Site B (with Advanced VM Support) i get an error which says all UPS have to be the same Type...

To buy a third 15kVA is not an Option. Besides the Money there is not enough room where i could install the UPS.

So your Solution will not work for me, unfortunatelly.

Any other ideas?

Thank you.

Regards

Marco

[BillP]

Marco,

What are the exact model numbers of your Smart-UPS? The information can be found on UPC label in each UPS.

Thanks,

[BillP · Accepted solution]

Hi Marco,

Is it possible to power the Storage Array in Site B using the 15kVA UPS?

NOTE: Since there is no physical UPS configured in the PCNS Setup in Site A it would be necessary to configure an event command file to trigger the shutdown of the physical server in Site A e.g.: using the native Windows shutdown.exe.

shutdown /s /f /t %delay% /d UP:6:12 /c "PowerChute Network Shutdown has initiated a system shutdown."

The command above will trigger shutdown on the physical server using a grace period (%delay%).

The value you set for %delay% above would need to be sufficient to allow VM migration/shutdown to complete as well as the ESXi host shutdown command to be sent.

[BillP]

Hi dgrehan,

thanks.

I was yesterday at the customers site and we did the same as you describe.

Your picture shows the correct situation for me. The storage is powered by the 15kVA on Site B.

I did the same you describe. Except my PCNS on Site A (with Singe UPS Configuration) is a Virtual Appliance, not a physical Host. Because its a "Single UPS Config" i thougt i dont need a physical Host there.

The physical PCNS Host on Site B is connected to his own UPS (Smart UPS 3000VA Rack) over USB. The physical PCNS with his own UPS has a longer remaining runtime than the other Ressources/UPS.

To Shutdown the physical PCNS Host gracefully we use the own Windows UPS Feature.

I see i did not need a own UPS for my physical PCNS Host if i work with a Shutdown-Script your posted. Thats right.

But we wanted to secure the Host seperatly to still monitor Site B if Site A was going down. In your setup the PCNS Host on Site B will also Shutdown, if there is a Shutdown triggered on Site A.

So we decided to use a own UPS with longer runtime.

So your Solution works for me, dgrehan.

Thanks you for your help. Thanks to Bill P., too.

Regards

Marco