Central Management for 2 Smart UPS

[Anonymous user]

Hopefully someone can clarify my confusion on what I really need. We have 2x SMX3000 each powering independent racks with independent equipment including VM environment. I'm looking for central management/monitoring for load etc., as well as be able to graceful shutdown equipment, particularly VMs. We'd like to be able to stagger the powering on of different port groups, so that things like our SAN comes up before our VM hosts. 

I looked at using Powerchute Network Shutdown, and it seems to do what we want, but it only seems to support one UPS. I can't find a way of adding another device, but I very likely could be missing something. 

I looked at the trial of StruxureWare, but I can't figure how to setup the graceful shutdowns, though I thought I read that it can do that. Maybe I missed something.

Any help on what makes the most sense for me would be appreciated. The big driver is just being able to get alerts on load, and be able to gracefully shut things down. 

Thanks in advance! 

[BillP]

Hi Dave,

At a high level, you'll want a Network Management Card (NMC) inside of each UPS. Then, any equipment you want to gracefully shut down, such as an ESXi host, you'd install PowerChute Network Shutdown (PCNS) on. Each PowerChute install would monitor the UPS status and help make sure the particular OS or server shuts down gracefully. PCNS and the NMC in each UPS work together to achieve graceful shutdown of attached loads and UPS for any given UPS and its load. You point PCNS to the IP address of the NMC that is installed in the UPS it is attached to and they share status updates with each other. There are also other specific settings you can do, such as shutting down certain devices earlier than others, load shedding, etc as you were talking about.

For monitoring of the UPS status of itself (outside of performing the graceful shutdown tasks) can be done from each NMC device. They support emails, SNMP, etc for monitoring the status of the UPS, load, outlet groups, battery status, etc. So, you can configure each NMC in the UPS to do that. We also offer tools that allow you to copy a configuration from one UPS/NMC combo to another to ease sharing of configuration settings such as email alerts. (I think PCNS also has a configuration file too these days you can share across different installs..).

StruxureWare Data Center Expert's purpose is to monitor and manage SNMP devices - APC devices such as UPSs (with NMCs installed) or third party devices. You can also monitor Modbus devices. It has a few special features for APC devices such as mass firmware upgrades and mass configuration. You can also configure alerts on it centrally instead of each NMC. This makes a little more sense if you have a fleet of UPSs, other data center equipment, SNMP devices, or third party devices that you want to graph temperatures, voltages, and similar over time. It also supports video surveillance for  APC NetBotz devices. It is ultimately one of the more powerful enterprise Data Center Infrastructure Monitoring tools out there that maybe you've heard of similar things. It also has functions of an SNMP Network Management System like Nagios or HP OpenView.

Thus, you could try the trial if you wanted and see what you think but it may not be cost effective for two UPSs. People typically monitor thousands of devices in here but it starts to become really useful and powerful when you have smaller amounts like tens of devices or hundreds of devices. StruxureWare does not do anything with graceful shutdown and is more to monitor status of SNMP devices, send alerts, etc.

Hope this helps! We have some white papers and stuff available for PowerChute configuration on VMware and with outlet groups on your UPS when you get to the nitty gritty of configuring things.

[Anonymous user]

Thanks Angela! I think that makes much more sense now. Our environment is a little bit odd, so bare with me while I work through this, and you can correct me if I misstep. As a note, both our UPSs already have NMC cards. 

We have 2 racks. Rack 1 is our typical production environment. Its a firewall, 3 switches, 2x ESXi hosts (managed via vCenter) with a NAS. I installed "PCNS for virtualization" as a virtual appliance on our ESXi, and it picked up vCenter and gave me that awesomeness. To address physical servers, it sounds like I'd need to install PCNS on each server? Can it shutdown equipment via SNMP? 

In Rack 2, we have a weird setup. Since this is a development environment, we have 5-6 standalone ESXi hosts. If I wanted to gracefully shut VMs down here, it sounds like I'd have to put PCNS on each host? 

If I'm understanding this right, there is no "central management", each server's shutdown behavior is individually configured. With that, is it possible to set the priority for Outlet Groups? So Group 1 shutdown immediately, Group 2 shutdown at 50% battery, etc.? Is it possible to do the same when power is restored, set Group 1 to come up immediately, Group 2 delay for 1 min, etc.?

Sorry if this is long winded and confusing. This isn't how I envisioned Powerchute functioning, so I'm trying to figure out how to lay this out. 

[BillP]

Hi Dave,

When we get to PowerChute setup, I need to ask my colleague  Bill to chime in. My expertise is the NMC itself so I am not as familiar with all of the ins and outs of PowerChute configurations.

Bill, can you comment here for Dave?

[BillP · Accepted solution]

Hi,

On 11/14/2017 4:40 PM, Dave said:

To address physical servers, it sounds like I'd need to install PCNS on each server?

Correct.

On 11/14/2017 4:40 PM, Dave said:

Can it shutdown equipment via SNMP? 

PCNS has the ability to trigger command files / scripts. PCNS would trigger the command file during a configured event such as on battery or when the OS is being command down. As example a command file has been written to power down RAID and is set to run after the UPS has been on battery for 10 minutes. The UPS reports it is on battery. PCNS waits 10 minutes and since the event has not cleared runs the command file. The command file has been written to send SNMP OID to the RAID. The RAID receives the SNMP OID and reacts. 

On 11/14/2017 4:40 PM, Dave said:

In Rack 2, we have a weird setup. Since this is a development environment, we have 5-6 standalone ESXi hosts. If I wanted to gracefully shut VMs down here, it sounds like I'd have to put PCNS on each host? 

Correct. You should install PCNS on each host. 

On 11/14/2017 4:40 PM, Dave said:

If I'm understanding this right, there is no "central management", each server's shutdown behavior is individually configured. With that, is it possible to set the priority for Outlet Groups? So Group 1 shutdown immediately, Group 2 shutdown at 50% battery, etc.? Is it possible to do the same when power is restored, set Group 1 to come up immediately, Group 2 delay for 1 min, etc.?

You can configure the shut down and start up of the outlet groups from within the NMC interface under Configuration - Outlet Groups. You would configure load shedding to power the outlet group down and configured the power on delay for the start up. Each outlet group can be configured individually. 

Example: Group  1 is set for load shedding after the UPS has been on battery for 600 seconds. When the delay is reached the NMC send a signal to any PCNS client that has been registered with that outlet group. PCNS starts the shutdown process. The Outlet group waits any power off delay and then cuts power to the outlets. The outlet group will wait until AC is restored and then wait the power on delay and or minimum return runtime and then restore AC to the outlets.