VMware vSphare Resource Allocation 

Describe Reservation Limit and share-

Reservations:- Reservations serve to act as guarantees of a particular resource.

Limits:- Limits are, quite simply, a way to restrict the amount of a given resource that a VM can use. Limits enforce an upper ceiling on the usage of resource.

Shares:- Shares serve to establish priority. Shares apply only during periods of host resource contention and serve to establish prioritized access to host resource.

Reservation-guaranted resource,

Limit-Upper lmit of given resource,

Share-Prioritize resource access in time of contension

What are the advance memory management techniques used by vSphare-

i) Transparent page sharing (TPS) ii) Ballooning iii) Swapping iv) Memory compression?

What is Transparent Page Sharing (TPS)- identical memory pages are shared among VMs to reduce the total number of memory pages needed. The hypervisor computes hashes of the contents of memory pages to identify pages that contain identical memory.

What is memory ballooning-

è'Ballooning' involves the use of a driver —called the balloon driver — installed into the guest OS.

èThis driver is part of VMware Tools installation.

è The balloon driver can respond to commands from the hypervisor to reclaim memory from that particular guest OS.

èThe balloon driver request memory from the guest OS — a process calling 'inflating'—and then passing that memory back to the hypervisor for use by other VMs.

èIf guest OS can give up pages it is no longer using then there may not be any performance degradation on the applications running inside that guest OS.

èIf the guest OS is already under memory pressure — then  “inflating” the balloon driver will invoke guest OS paging (or Guest swapping), which will degrade performance.

Describe briefly how Balloon driver works-

èThe balloon driver is part of the VMware Tools.

è>As such, it is a guest OS– specific driver, meaning that Linux VMs would have a Linux-based balloon driver, Windows VMs would have a Windows-based balloon driver, and so forth.

èRegardless of the guest OS, the balloon driver works in the same fashion. When the ESXi host is running low on physical memory, the hypervisor will signal the balloon driver to grow. To do this, the balloon driver will request memory from the guest OS. This causes the balloon driver’s memory footprint to grow, or to inflate. The memory that is granted to the balloon driver is then passed back to the hypervisor. The hypervisor can use these memory pages to supply memory for other VMs, reducing the need to swap and minimizing the performance impact of the memory constraints.

èWhen the memory pressure on the Esxi host passes, the balloon driver will deflate, or return memory to the guest OS.

The key advantage that ESXi gains from using a guest-OS-specific balloon driver in this fashion is that it allows the guest OS to make the decision about which pages can be given to the balloon driver process (and thus released to the hypervisor). In some cases, the inflation of the balloon driver can release memory back to the hypervisor without any degradation of VM performance because the guest OS is able to give the balloon driver unused or idle pages.

What is hypervisor swapping?

èGuest OS swapping falls strictly under the control of the guest OS and is not controlled by the hypervisor.

èIf none of the previously described technologies trim guest OS memory usage enough, the ESXi host will swap memory pages out to disk in order to reclaim memory that is needed elsewhere.

è ESXi’s swapping takes place without any regard to the guest OS. As a result guest OS performance is severely impacted if hypervisor swapping is invoked.

What is memory compression-?

When an ESXi host gets to the point that hypervisor swapping is necessary, then VMkernel will attempt to compress memory pages and keep them in RAM in a compressed memory cache. Pages that can be successfully compressed by at least 50 percent are put into the compressed memory cache instead of being written to disk and can then be recovered much more quickly if the guest OS needs that memory page. Memory Compression is invoked only when the ESXi host reaches to the point where swapping is needed.

From where a VM without Reservation gets is memory from-

What is VMkarnel swap-

When an ESXi host doesn’t have enough RAM available to satisfy the memory needs of the VMs it is hosting and when other technologies such as transparent page sharing, the balloon driver, and memory compression aren’t enough, then VMkernel is forced to page some of each VM’s memory out to the individual VM’s VMkernel swap file.

VMkarnel swap:-

VMkernel swap is actually the hypervisor swapping mechanism. VMkernel swap is implemented as a file with a .vswp extension that is created when a VM is powered on. These per-VM swap files created by the VMkernel reside, by default, in the same datastore location as the VM’s configuration file (.VMX) and virtual disk files (.VMDK) (although you do have the option of relocating the VMkernel swap).

 In the absence of a memory reservation — the default configuration is — this file will be equal in size to the amount of RAM configured for the VM. Thus, a VM configured for 4 GB of RAM will have a VMkernel swap file that is also 4 GB in size and stored, by default, in the same location as the VM’s configuration and virtual disk files.

In theory, this means a VM could get its memory allocation entirely from Hypervisor's physical memory or VMkernel swap ie. from disk. If VMkarnel swap memory is assigned then some performance degradation for VM is obvious because disk access time is several orders of magnitude slower than RAM access time.

Configured memory – memory reservation =Expected size of swap file (.vswp)

Do "Transparent Page Sharing (TPS)" works for ‘Reserved Memory’- Yes

What about "Memory Ballooning"-No

Reserved memory can be shared via transparent page sharing (TPS). Transparent page sharing does not affect the availability of reserved memory because the shared page is still accessible to the VM.

What is LIMIT?  What are its impact on guest OS-

It sets the actual limit or upper ceiling on how much allocated physical resource may be utilized by that VM.

The key problem with the use of memory limits is that they are enforced by vmkernel without any guest OS or VM awareness. The guest OS will continue to behave as if it can use allocated maximum RAM, completely unaware of the limit that has been placed on it by the hypervisor. Setting a memory limit will have a significant impact on the performance of the VM because- as a result the guest OS will constantly be forced to swap pages to disk (guest OS swapping, not hypervisor swapping).

Why use memory limit-

Use memory limits as a temporary measure to reduce physical memory usage in your ESXi hosts.

Perhaps you need to perform troubleshooting on an ESXi host and You want to temporarily push down memory usage on less-important VMs so that you don’t overcommit memory too heavily and negatively impact lots of VMs. Limits would help in this situation.

In general, then, you should consider memory limits as a temporary stop-gap measure when you need to reduce physical memory usage on an ESXi host and a negative impact to performance is acceptable.

                       CPU Utilization

Like shares, reservations, and limits, what is the fourth option available for managing CPU utilization?

CPU affinity. CPU affinity allows an administrator to statically associate a VM to a specific physical CPU core. CPU affinity is generally not recommended; it has a list of rather significant drawbacks:

è CPU affinity breaks vMotion.

è Because vMotion is broken, you cannot use CPU affinities in a cluster where vSphere DRS isn’t set to Manual operation.

è The hypervisor is unable to load-balance the VM across all the processing cores in the server. This prevents the hypervisor’s scheduling engine from making the most efficient use of the host’s resources.

Remember:- We  use CPU Reservation, Limit and Share to control CPU clock cycle allocation (Core speed).

What is the difference between Memory Reservation and CPU Reservation-

A CPU Reservation is very different than a Memory Reservation when it comes to “sharing” reserved CPU cycles. Reserved Memory, once allocated to the VM, is never reclaimed, paged out to disk, or shared in any way. The same is not true of CPU Reservations.

The ESXI host has two idle VMs running. The shares are set at the defaults for the running VMs. Will the Shares values have any effect in this scenario-?

No. There’s no competition between VMs for CPU time because both are idle. Share comes in to play in time of resource contention.

The ESX host with dual, single-core, 3 GHz CPUs has two equally busy VMs running (both requesting maximum CPU capacity). The shares are set at the defaults for the running VMs. Will the Shares values have any effect in this scenario-?

No. Again, there’s no competition between VMs for CPU time, this time because each VM is serviced by a different core in the host.

Remember:-CPU Affinity Not Available with Fully Automatic DRS enabled Clusters.

                                                               

If you are using a VSphere Distributed Resource Scheduler–enabled cluster configured in fully automated mode, CPU affinity cannot be set for VMs in that cluster. You must configure the cluster for manual or partially automated mode in order to use CPU affinity.

Describe CPU Reservation, Limit and Share?

è Reservations set on CPU cycles provide guaranteed processing power for VMs. Unlike memory, reserved CPU cycles can and will be used by ESXi to service other requests when needed. As with memory, the ESXi host must have enough real, physical CPU capacity to satisfy a reservation in order to power on a VM.

è Limits on CPU usage simply prevent a VM from gaining access to additional CPU cycles even if CPU cycles are available to use. Even if the host has plenty of CPU processing power available to use, a VM with a CPU limit will not be permitted to use more CPU cycles than specified in the limit. Depending on the guest OS and the applications, this might or might not have an adverse effect on performance.

è Shares are used to determine CPU clock cycle allocation when the ESXi host is experiencing CPU contention. CPU shares grant CPU access on a percentage basis calculated on the number of shares granted out of the total number of shares assigned.

What is Resource Pool? Why it is required?

Resource pool basically is a special type of container object, much like a folder, mainly used to group VM's with similar resource allocation needs to avoid administrative overhead. Resource pool uses reservations, limits, and shares to control and modify resource allocation behavior, but only for memory and CPU.

What is Expandable Reservation in resource Pool-

A Resource Pool provides resources to its child objects. A child object can either be a virtual machine or another resource pool. This is what called the parent-child relationship.

But what happens if the child objects in the resource pool are configured with reservations that exceeds the reservation set on the parent resource pool-- in that case parent resource pool needs to request protected resources from its parent resource pool. This can only be done if expandable reservation is enabled.
Please note that the resource pool request protected or reserved resources from its parent resource pool, it will not accept resources that are not protected by a reservation.

You want to a understand Resource Pool's resource allocation, from where you can see allocation of resources to objects within the vCenter Server hierarchy-

Clusters "Resource Allocation" tab can verify the allocation of resources to objects within the vCenter Server hierarchy.

Remember:-Shares Apply Only During Actual Resource Contention-

Remember that share allocations come into play only when VMs are fighting one another for a resource — in other words, when an ESXi host is actually unable to satisfy all the requests for a particular resource. If an ESXi host is running only eight VMs on top of two quad-core processors, there won’t be contention to manage (assuming these VMs have only a single vCPU and Shares values won’t apply.

What is Processor core? Thread?  what is Hyperthreading? what is Logical CPU and Virtual CPU-

Processor : responsible of all processing operations. We define multiple processors in a server by Socket.

Core : one operations unit or processing unit Inside your physical processor.

Hyper-Threading:- Normally a processor Core can handle one thread or one operation at a same time (time means processor time slot). But with Hyper-Threading enabled a processor Core can handle two threads at the same time.

Logical Processor : The number of threads that processor Cores can handle in a machine is the number of logical processor. So if you want to know how much logical processor do you have, just count the total number of threads a processor can handle.

How much logical processor do you have:-

Cores Count = Processor Count X CoresCountPerProcessor

Logical Processor Count = CoresCount X ThreadCount

so

No-of-Processor-(Socket) X Cores-Per-Processor X ThreadCount = Logical Processor Count

Virtual Processor:  when you create a virtual machine you do assign to it a processor. Like vRAM, VHD, Virtual network interface, we can assign a Virtual Processor (VP) to a virtual machine. Actually virtual Processor is nothing but the physical processor’s TimeSlot slice that will be given to the virtual machine.

What is SMP? Symmetric Multiprocessing: SMP is processing of a program by multiple processors that share a common operating system and memory.

What are the  maximum Virtual CPU Limitations of VMware?

Maximum number of virtual CPUs in a VM depends on the following:-

(a) Number of logical CPUs on the host,

(b)The ESXi host license,

(c)Type of guest operating system is installed on VM. 

What is Network Resource Pool?

A 'network resource pool' is a resource pool which control network utilization. A network resource pool can control outgoing network traffic or outgoing network bandwidth with the help of shares and limits. This feature is referred to as vSphere Network I/O Control (NetIOC).

Network resource pool or NetIOC’s disadvantage:-

(X)Controls outgoing Traffic Only

(Y)Works only on a Distributed Switch

What is System Network Resource Pool? What is Custom Resource Pool-

When you enable vSphere NetIOC, vSphere activates six predefined system network resource pools:

l Fault Tolerance (FT) Traffic

l Virtual Machine Traffic

l vMotion Traffic

l Management Traffic

l iSCSI Traffic

l NFS Traffic

Custom Resource Pool is used to control customer’s resource utilization?

Remember:- You Can’t Map Port Groups to System defined resource Pools

Port groups can only be mapped to user-defined network resource pools, not system network resource pools. 

How do you enable NetIOC?

First, you must enable NetIOC on that particular vDS. Second, you must create and configure custom network resource pools as necessary.

What are three basic settings each network resource pool consists of?

'Physical Adapter Shares'

'Host Limit'

QoS Priority Tag

èPhysical Adapter Shares'- priority for access to the physical network adapters when there is network contention.

è  'Host Limit'- upper limit on the amount of network traffic a network resource pool is allowed to consume (in Mbps)

è QoS Priority Tag- The QoS (Quality of Service) priority tag is an 802.1p tag that is applied to all outgoing packets. Configured upstream network switches can further enhance and enforce the QoS just beyond the ESXi host.

What are the pre-requisites of storage I/O control (SIOC)-

SIOC has a few requirements you must meet:

(i)All SIOC enabled Datastores should be under a single vCenter Server-

(j)No RDM Support, NO NFS Support

(k) Datastore with Multiple Extents Support not supported- 

Remember:- check Array Auto-Tiring compatibility with SIOC

 What is auto-tiering — the ability for the array to seamlessly and transparently migrate data between different storage tiers (SSD, FC, SAS, SATA).

How do you Enabling Storage I/O Control-

First, enable SIOC on one or more datastores.

Second, assign shares or limits to storage I/O resources on individual VMs.

Remember these points about SIOC-

SIOC is enabled on a per-datastore basis.

By default, SIOC is disabled for a datastore, meaning that you have to explicitly enabled SIOC if you want to take advantage of its functionality.

While SIOC is disabled by default for individual datastores, it is enabled by default for Storage DRS–enabled datastore clusters that have I/O metrics enabled for Storage DRS.

How Storage I/O control SIOC works-

SIOC uses disk latency as the threshold to enforce Shares values-

 SIOC uses disk latency as the threshold to determine when it should activate and enforce Shares values for access to storage I/O resources. Specifically, when vSphere detects latency is in excess of a specific threshold value (measured in milliseconds), SIOC is activated.

For controlling the use of storage I/O by VMs SIOC uses shares and limits-

SIOC provides two mechanisms for controlling the use of storage I/O by VMs: shares and limits.  the Shares value establishes a relative priority as a ratio of the total number of shares assigned, while the Limit value defines the upper ceiling on the number of I/O operations per second (IOPS) that a given VM may generate.

To guarantee certain levels of performance, your IT director believes that all VMs must be configured with at least 8 GB of RAM. However, you know that many of your applications rarely use this much memory. What might be an acceptable compromise to help ensure performance?

One way would be to configure the VMs with 8 GB of RAM and specify a reservation of only 2 GB. VMware ESXi will guarantee that every VM will get 2 GB of RAM, including preventing additional VMs from being powered on if there isn’t enough RAM to guarantee 2 GB of RAM to that new VM. However, the RAM greater than 2 GB is not guaranteed and, if it is not being used, will be reclaimed by the host for use elsewhere. If plenty of memory is available to the host, the ESXi host will grant what is requested; otherwise, it will arbitrate the allocation of that memory according to the shares values of the VMs.

A fellow VMware administrator is a bit concerned about the use of CPU reservations. She is worried that using CPU reservations will “strand” CPU resources, preventing those reserved but unused resources from being used by other VMs. Are this administrator’s concerns well founded?

For CPU reservations- No. While it is true that VMware must have enough unreserved CPU capacity to satisfy a CPU reservation when a VM is powered on, reserved CPU capacity is not “locked” to a VM like memory reservation. If a VM has reserved but unused CPU capacity, that capacity can and will be used by other VMs on the same host. The other administrator’s concerns could be valid, however, for memory reservations.

Your company runs both test/development workloads and production workloads on the same hardware. How can you help ensure that test/development workloads do not consume too many resources and impact the performance of production workloads?

Create a resource pool and place all the test/development VMs in that resource pool. Configure the resource pool to have a CPU limit and a lower CPU shares value. This ensures that the test/development will never consume more CPU time than specified in the limit and that, in times of CPU contention; the test/development environment will have a lower priority on the CPU than production workloads.

Name two limitations of Network I/O Control-

NIOC works only with

è(i) vSphere Distributed Switches,

è(ii)The ability to only control outbound network traffic

è(iii) The fact that it requires vCenter Server in order to operate

è (iv)”System network resource pools” cannot be assigned to user-created port groups?

What are the requirements for using Storage I/O Control-

èAll datastores and ESXi hosts that will participate in Storage I/O Control must be managed by the same vCenter Server instance.

èRaw Device Mappings (RDMs) and NFS datastores are not supported for SIOC.

èDatastores must have only a single extent; datastores with multiple extents are not supported?