Your Daily VMware quiz!

tomtom1

- The maximum RAM for a vSphere 5.5 VM is 2 TB. True/False?
No, still 1TB (assuming by a vSphere 5.5 VM - You mean HW version 10)

- If an admin were to upgrade a VMFS 3 datastore to a VMFS 5 datastore, the block size of the datastore changes to 1 MB. True/False?
No, block size on an upgraded datastore is retained.

- A customer has purchased ESXi Foundations and is thinking of using host profiles for ease of management. Host profiles will be available to them. True/False?
If you mean the acceleration kits, answer is no.

- vCenter 5.5 can manage ESXi 5.x hosts. True/False?
True

- Changing the network label of a VM from "ProdVMs" to "FinanceVMs" will not cause a slight outage to the VM. True/False
Had to test this one out, result was false.

- You can vMotion across clusters if the networking is setup correctly. True/False?
True

- You like this quiz!. True/False?
True

Essendon

2 more questions could do with an answer! Anyone?

Asif Dasl

I believe tomtom1 got the rest right.

- ESXi stands for Elastic Sky X integrated. True/False? (I know this is a trivia question!)
True

- The "host-only" Networking setting in VMware Workstation will allow VM's to get out to the Internet. True/False?
False

- You like this quiz!. True/False?
True

kj0

Everything is False!



How'd I do?

tomtom1

kj0 wrote: »

Everything is False!



How'd I do?

I'd go for the VCP-DCV exam right now!

kj0

tomtom1 wrote: »

I'd go for the VCP-DCV exam right now!

Done.


- The maximum RAM for a vSphere 5.5 VM is 2 TB. True/False?
False

- If an admin were to upgrade a VMFS 3 datastore to a VMFS 5 datastore, the block size of the datastore changes to 1 MB. True/False?
False
- ESXi stands for Elastic Sky X integrated. True/False? (I know this is a trivia question!)
False - Elenor's Super Extreme Intelligence. (True for those who don't know what to think)
- A customer has purchased ESXi Foundations and is thinking of using host profiles for ease of management. Host profiles will be available to them. True/False?
False - enterprise.
- The "host-only" Networking setting in VMware Workstation will allow VM's to get out to the Internet. True/False?
False
- vCenter 5.5 can manage ESXi 5.x hosts. True/False?
True.
- Changing the network label of a VM from "ProdVMs" to "FinanceVMs" will not cause a slight outage to the VM. True/False?
True
- You can vMotion across clusters if the networking is setup correctly. True/False?
"IF" being key. True.
- You like this quiz!. True/False?
Check yesterdays tweet.

tomtom1

kj0 wrote: »

- A customer has purchased ESXi Foundations and is thinking of using host profiles for ease of management. Host profiles will be available to them. True/False?
False - enterprise.

Wrong, enterprise plus

jibbajabba

tomtom1 wrote: »

Wrong, enterprise plus

Correct

Some VMware blurp wrote:

To use the Host Profiles feature, upgrade to VMware vSphere 4.1 or 5.x Enterprise Plus Edition. To see a complete list of features shipped in VMware vSphere 4.1 and 5.x Enterprise Plus Edition, see the VMware Store.

Essendon

Question 5 (one or more choices may be right)

Which of the following will need a reboot of your ESXi 5.x host?

1. Adding a uplink to the management network of your host
2. Adding of the host to a distributed vSwitch
3. Modifying the number of ports on a vSwitch
4. Modifying the name of a port group
5. Attaching a host profile to it

Answer: 3 and 5 (like tomtom1 said, 5 may or may not apply depending on the setting in the host profile but for this question 5 is also a correct answer)

Memory reservation set on VM will

1. Guarantee a minimum to the VM during times of contention on the host
2. Guarantee a minimum to the VM at all times
3. Cause a 50% reduction in the size of the VM's .vswp file
4. Need a reboot of the VM to be enforced
5. Decrease the amount of memory available to other VM's

Answer: 2, 4 and 5. Reservations guarantee a minimum at all times and are activated only when a VM boots up but allocated only as needed. Because reservations are a guaranteed minimum at all times, this memory is not available to other VM's even if there's contention. Reservations MUST be sparingly otherwise you may see other VM's suffer.

You'd install VMware Tools on a VM so that

1. You can restart the guest OS from the vSphere Client
2. Enhance its graphics performance
3. You can sync its time with the host (question within a question - will you recommend this in an AD domain environment)
4. VMware HA VM monitoring will work
5. You can thin provision the VM and be able to vMotion it

Answer: 1, 2, 3 and 4. I wouldnt use VMware Tools' time sync, let your AD PDC do its thing. Check here for more info. Use one or other, not both. HA VM monitoring needs VMware Tools to heartbeat for monitoring to work. If it doesnt receive the heartbeat within a certain period of time then action to restart the VM can be taken at certain intervals and a number of times depending on your settings.

Remember you got to tell me why a choice or choices are correct!

tomtom1

Essendon wrote: »

Question 5 (one or more choices may be right)

Which of the following will need a reboot of your ESXi 5.x host?

1. Adding a uplink to the management network of your host
2. Adding of the host to a distributed vSwitch
3. Modifying the number of ports on a vSwitch
4. Modifying the name of a port group
5. Attaching a host profile to it

Memory reservation set on VM will

1. Guarantee a minimum to the VM during times of contention on the host
2. Guarantee a minimum to the VM at all times
3. Cause a 50% reduction in the size of the VM's .vswp file
4. Need a reboot of the VM to be enforced
5. Decrease the amount of memory available to other VM's

You'd install VMware Tools on a VM so that

1. You can restart the guest OS from the vSphere Client
2. Enhance its graphics performance
3. You can sync its time with the host (question within a question - will you recommend this in an AD domain environment)
4. VMware HA VM monitoring will work
5. You can thin provision the VM and be able to vMotion it

Remember you got to tell me why a choice or choices are correct!

1. Adding a uplink to the management network of your host
No
2. Adding of the host to a distributed vSwitch
No
3. Modifying the number of ports on a vSwitch
Yes
4. Modifying the name of a port group
No
5. Attaching a host profile to it
Depends on the specific settings that are being pushed down by the HP but, for arguments sake, yes.

I'll leave the other 2 open for somebody else

kj0

Essendon wrote: »

Question 5 (one or more choices may be right)

Which of the following will need a reboot of your ESXi 5.x host?

1. Adding a uplink to the management network of your host
2. Adding of the host to a distributed vSwitch
3. Modifying the number of ports on a vSwitch
4. Modifying the name of a port group
5. Attaching a host profile to it

Memory reservation set on VM will

1. Guarantee a minimum to the VM during times of contention on the host
2. Guarantee a minimum to the VM at all times
3. Cause a 50% reduction in the size of the VM's .vswp file
4. Need a reboot of the VM to be enforced
5. Decrease the amount of memory available to other VM's

You'd install VMware Tools on a VM so that

1. You can restart the guest OS from the vSphere Client
2. Enhance its graphics performance
3. You can sync its time with the host (question within a question - will you recommend this in an AD domain environment)
4. VMware HA VM monitoring will work
5. You can thin provision the VM and be able to vMotion it

Remember you got to tell me why a choice or choices are correct!

I'll Explain my answers once I get a moment, but got to get in quick.

tomtom1

To keep this topic going, Essendon and myself will be posting questions here.

Your company is in the process of upgrading their datacenter infrastructure. For the compute section of the datacenter (CPU and memory), new servers have been bought. Instead of going with 1 GbE networking, the servers have 2 10 GbE connections. You manager worries about the fact that since there is no physical separation and limitation of network traffic, one traffic flow might be able to use up all of the bandwidth. How would you tackle this?


You want to make sure that the datastore you use for virtual machines matches the SLA, which is either Gold or Silver. If you create a new virtual machine, you would like to make sure that it preselects the right datastores based on the SLA level you chose. Tell me what to configure.

Good luck!

tstrip007

1. Configure traffic shaping or NIOC? and I believe a distributed switch will allow for more granular rules and policies 2. Create storage profiles

Essendon

Great questions @tomtom1

Yep, you'd create a distributed switch and add the hosts to the switch. Then you'd enable NIOC on the vDS. NIOC lets you have different workloads coexist on the same pipe without strangling out other types. vMotion is a particular culprit, being a bursty mechanism it will take up the entire bandwidth causing other types of traffic to suffer. To ensure this doesnt happen, you should enable NIOC on the vDS and allocate shares to each type of traffic as is suitable for your installation. One thing to note though is the default system defined resource pools apply to the entire switch, but if you were to create your own resource pools, you can set them on port groups of your choice.

Profile driven storage is the way to go here. Though I've found in my experience they can be difficult to enforce, especially when people provision a new .vmdk to VM's. Some admins dont seem to care where the vmdk goes. I am not sure if there's an alarm that can be configured to alert you to the fact that a VM's disks arent compliant with its storage profile. Maybe there's some PowerShell magic that can be used for this.

It's important to note that VASA doesnt provide any performance data to StorageDRS, some people tend to think otherwise. Storage DRS uses performance data from SIOC to do its thing.

Essendon

Question 7

A customer has asked hired you onto begin looking at virtualizing their physical server. In order to gauge the infrastructure and check the workloads, which tool would you employ?

1. VMware Chargeback Manager
2. VMware Capacity Analyzer
3. VMware Infrastructure Navigator
4. VMware vCenter Heartbeat
5. VMware Operations Manager


A Systems Manager walks upto you and whinges that his application isnt performing well. You check that the application is on a VM with 2 vCPU's and 8GB configured memory and Task Manager shows a mere 30% memory utilization. What should you check to determine the cause of the problem? The application also has an SQL backend (on a different VM).

You have hired a junior virtualization administrator and all's well for a few days until one fine morning you ask her to configure Linked Mode for 2 vCenters. Though these vCenters are for the Prod and DR sites, company policy states that admins on one infrastructure not be able to muck around with the other infrastructure. The next day you find out that admins from the DR site have created resource pools in the Prod site. What's caused this and how will you undo it?

tomtom1

Come on guys, nobody?

Essendon

The weekend's the culprit!

Anyways, +rep on offer if anyone provides a decent, well-explained answer to Question 7. Well explained, right?!

QHalo

Essendon wrote: »

Question 7

A customer has asked hired you onto begin looking at virtualizing their physical server. In order to gauge the infrastructure and check the workloads, which tool would you employ?

1. VMware Chargeback Manager
2. VMware Capacity Analyzer
3. VMware Infrastructure Navigator
4. VMware vCenter Heartbeat
5. VMware Operations Manager


A Systems Manager walks upto you and whinges that his application isnt performing well. You check that the application is on a VM with 2 vCPU's and 8GB configured memory and Task Manager shows a mere 30% memory utilization. What should you check to determine the cause of the problem? The application also has an SQL backend (on a different VM).

You have hired a junior virtualization administrator and all's well for a few days until one fine morning you ask her to configure Linked Mode for 2 vCenters. Though these vCenters are for the Prod and DR sites, company policy states that admins on one infrastructure not be able to muck around with the other infrastructure. The next day you find out that admins from the DR site have created resource pools in the Prod site. What's caused this and how will you undo it?

If you want to check current workloads that are already virtualized, you can go ahead and use vCOPs. If you're wanting to look at physical and more than one and you're a partner then use Capacity Analyzer. But if you're a customer then use perfmon tools to look at the server in question for a period of time.

Start with what has changed recently on the server? Has it always behaved this way? Is this something new? Can you give me a time period that this happens or does it happen all the time? Then look at the host statistics in vCenter. Is the host over-consumed? Are there any resource pools or reservations in place on the VM? Then I would look at is %RDY and %CSTP of the VM to determine if there's a ready issue. Then move to DAVG, KAVG and GAVG to determine if the back-end storage is causing any latency. After looking at all those spots I would move onto the SQL server and go through the same steps.

Linked mode grants RBAC access to both vCenters. If you don't want someone doing something on either side, then restrict access on the side you don't want them to do things on via the vCenter that runs that infrastructure.

Essendon

Great reply there QHalo. I'd like to point how he's answered questions 2 and 3. See there's a nice flow of thought, he starts off with the basics and then progresses to think outside the box. In my experience, complaints like these usually arise when mass batch jobs are being processed to the database. Co-locating the database VM and the app VM on the same host and/or same datastore can help fix the issue. But you must follow QHalo's thought process and then look at vMotioning/SvMotioning the VM's.

%RDY and %CSTP are of particular help in figuring out the cause of issues like this. But take care how you interpret them. Without having to rehash all the explanation, check this article > CPU Ready Time in VMware and How to Interpret its Real Meaning - Jonathan Kehayias .

Another one's %USED value, people tend to think a high %USED value is bad. No, it isnt bad. It just means that the VM is using it's vCPU(s) and actually putting it to work. It might even mean you need an extra vCPU to give to the VM. But only allocate an extra vCPU or two to a VM if absolutely needed. If the application that runs on the VM isnt multi-threaded, it will not benefit from more than 1 vCPU's.

While I'm at it, I'd like to add you need to look at the sizing of your NUMA nodes before you go about designing your VM infrastructure. You want your VM's to neatly fit into a NUMA node for compute and memory locality. Also that vNUMA is enabled by default for VM's larger than 8 vCPU's. "Larger than" is of importance here, because generally greater than 8 vCPU VM's are the ones that will not fit into a single NUMA node. Hence the need for VMware to allow vNUMA to the guest OS so that the guest OS can make its own decisions because it knows waaaay more about its workload than ESXi and make intelligent decisions on where to go.

QHalo

It's funny, in my original reply I had the multi-threaded portion in there as well but took it out for some reason. Probably due to poor editing when I was typing out my thoughts. Either way, always check your applications to ensure they're actually taking advantage of multi-threading. As for NUMA nodes, here's an example. If your VM needs 16 vCPUs (I have one right now that will eat all 16 when I give it them) and your server is a dual 8-core system, change your VM to 2 sockets, 8 cores. That should take care of vNUMA for you. Also, you want to make sure that you're also not breaking the NUMA boundary when it comes to RAM as well. If you only have 96GB of RAM in your system, a >48GB VM will break the NUMA boundary.

Essendon

Oh yes, another reason why you should right size your VMs! Watch out for both the CPU and RAM when it comes to NUMA config, either can break the NUMA boundary and performance will suffer. I've seen some people just disabling NUMA to get rid of NUMA problems, but they should've planned their hosts better to begin with.

And guys there is a difference in performance when you choose 2 sockets and 8 cores over 2 cores and 8 sockets depending on your NUMA node config.

tomtom1

Ok, next one:

Company policy states that all management traffic for the vSphere environment must be separated in it's own VLAN. The networking team has provided you with that VLAN, put it on the correct links and gave you the subnet information. During the configuration of the management VMkernel port, you come to the conclusion that by design, the default gateway for your subnet does not ping back. Your hosts are in an HA enabled cluster. What problems could this give you, and how would you go about mitigating them?

jibbajabba

By default HA needs to ping the gateway to determine whether a host is isolated or not. To fix it either tell your network team to get their act together (accompanied by donuts) or change the advanced settings of HA (by using das.isolationaddress) to use a different IP.

PS: VMware HA isolation failover would occur if the GW is not pingable.

tomtom1

jibbajabba wrote: »

By default HA needs to ping the gateway to determine whether a host is isolated or not. To fix it either tell your network team to get their act together (accompanied by donuts) or change the advanced settings of HA (by using das.isolationaddress) to use a different IP.

PS: VMware HA isolation failover would occur if the GW is not pingable.

If you change the advanced settings, I'm still missing one more advanced setting.

jibbajabba

Above is assuming you use an IP in the same range as your management network. If that is not the case then

das.allowNetworkX

To configure a different network used to check the isolation address.

tomtom1

jibbajabba wrote: »

Above is assuming you use an IP in the same range as your management network. If that is not the case then

das.allowNetworkX

To configure a different network used to check the isolation address.

Yes, I'm using an IP in the same subnet, hence the default gateway, that is almost always in the same segment.

Not what I am looking for, let me give you a slight hint: The default gateway is already set up on the ESXi system to the non pingable gateway the network team provided.

Essendon

I think he's after das.defaultisolationaddress=false and then have the other advanced setting.

tomtom1

Essendon wrote: »

I think he's after das.defaultisolationaddress=false and then have the other advanced setting.

jibbajabba

I thought das.defaultisolationaddress=false is assumed when using das.isolationaddress

tomtom1

jibbajabba wrote: »

I thought das.defaultisolationaddress=false is assumed when using das.isolationaddress

According to a few KB articles and the HA product documentation you have to specify it manually, and you need to do so in production. That's what I was after, otherwise good post.