Example Architectural Decision – ESXi Host Hardware Sizing (Example 1)

Problem Statement

What is the most suitable hardware specifications for this environments ESXi hosts?

Requirements

1. Support Virtual Machines of up to 16 vCPUs and 256GB RAM
2. Achieve up to 400% CPU overcommitment
3. Achieve up to 150% RAM overcommitment
4. Ensure cluster performance is both consistent & maximized
5. Support IP based storage (NFS & iSCSI)
6. The average VM size is 1vCPU / 4GB RAM
7. Cluster must support approx 1000 average size Virtual machines day 1
8. The solution should be scalable beyond 1000 VMs (Future-Proofing)
9. N+2 redundancy

Assumptions

1. vSphere 5.0 or later
2. vSphere Enterprise Plus licensing (to support Network I/O Control)
3. VMs range from Business Critical Application (BCAs) to non critical servers
4. Software licensing for applications being hosted in the environment are based on per vCPU OR per host where DRS “Must” rules can be used to isolate VMs to licensed ESXi hosts

Constraints

1. None

Motivation

1. Create a Scalable solution
2. Ensure high performance
3. Minimize HA overhead
4. Maximize flexibility

Architectural Decision

Use Two Socket Servers w/ >= 8 cores per socket with HT support (16 physical cores / 32 logical cores) , 256GB Ram , 2 x 10GB NICs

Justification

1. Two socket 8 core (or greater) CPUs with Hyper threading will provide flexibility for CPU scheduling of large numbers of diverse (vCPU sized) VMs to minimize CPU Ready (contention)

2. Using Two Socket servers of the proposed specification will support the required 1000 average sized VMs with 18 hosts with 11% reserved for HA to meet the required N+2 redundancy.

3. A cluster size of 18 hosts will deliver excellent cluster (DRS) efficiency / flexibility with minimal overhead for HA (Only 11%) thus ensuring cluster performance is both consistent & maximized.

4. The cluster can be expanded with up to 14 more hosts (to the 32 host cluster limit) in the event the average VM size is greater than anticipated or the customer experiences growth

5. Having 2 x 10GB connections should comfortably support the IP Storage / vMotion / FT and network data with minimal possibility of contention. In the event of contention Network I/O Control will be configured to minimize any impact (see Example VMware vNetworking Design w/ 2 x 10GB NICs)

6. RAM is one of the most common bottlenecks in a virtual environment, with 16 physical cores and 256GB RAM this equates to 16GB of RAM per physical core. For the average sized VM (1vCPU / 4GB RAM) this meets the CPU overcommitment target (up to 400%) with no RAM overcommitment to minimize the chance of RAM becoming the bottleneck

7. In the event of a host failure, the number of Virtual machines impacted will be up to 64 (based on the assumed average size VM) which is minimal when compared to a Four Socket ESXi host which would see 128 VMs impacted by a single host outage

8. If using Four socket ESXi hosts the cluster size would be approx 10 hosts and would require 20% of cluster resources would have to be reserved for HA to meet the N+2 redundancy requirement. This cluster size is less efficient from a DRS perspective and the HA overhead would equate to higher CapEx and as a result lower the ROI

9. The solution supports Virtual machines of up to 16 vCPUs and 256GB RAM although this size VM would be discouraged in favour of a scale out approach (where possible)

10. The cluster aligns with a virtualization friendly “Scale out” methodology

11. Using smaller hosts (either single socket, or less cores per socket) would not meet the requirement to support supports Virtual machines of up to 16 vCPUs and 256GB RAM , would likely require multiple clusters and require additional 10GB and 1GB cabling as compared to the Two Socket configuration

12. The two socket configuration allows the cluster to be scaled (expanded) at a very granular level (if required) to reduce CapEx expenditure and minimize waste/unused cluster capacity by adding larger hosts

13. Enabling features such as Distributed Power Management (DPM) are more attractive and lower risk for larger clusters and may result in lower environmental costs (ie: Power / Cooling)

Alternatives

1.  Use Four Socket Servers w/ >= 8 cores per socket , 512GB Ram , 4 x 10GB NICs
2.  Use Single Socket Servers w/ >= 8 cores , 128GB Ram , 2 x 10GB NICs
3. Use Two Socket Servers w/ >= 8 cores , 512GB Ram , 2 x 10GB NICs
4. Use Two Socket Servers w/ >= 8 cores , 384GB Ram , 2 x 10GB NICs
5. Have two clusters of 9 hosts with the recommended hardware specifications

Implications

1. Additional IP addresses for ESXi Management, vMotion, FT & Out of band management will be required as compared to a solution using larger hosts

2. Additional out of band management cabling will be required as compared to a solution using larger hosts

Related Articles

1. Example Architectural Decision – Network I/O Control for ESXi Host using IP Storage (4 x 10 GB NICs)

2. Example VMware vNetworking Design w/ 2 x 10GB NICs

3. Network I/O Control Shares/Limits for ESXi Host using IP Storage

4. VMware Clusters – Scale up for Scale out?

5. Jumbo Frames for IP Storage (Do not use Jumbo Frames)

6. Jumbo Frames for IP Storage (Use Jumbo Frames)

CloudXClogo

 

Example Architectural Decision – VMware HA – Percentage of Cluster resources reserved for HA

Problem Statement

The decision has been made to use “Percentage of cluster resources reserved for HA” admission control setting, and use Strict admission control to ensure the N+1 minimum redundancy level is maintained. However, as most virtual machines do not use  “Reservations” for CPU and/or Memory, the default reservation is only 32Mhz and 0MB+overhead for a virtual machine. In the event of a failure, this level of resources is unlikely to provide sufficient compute to operate production workloads. How can the environment be configured to ensure a minimum level of performance is guaranteed in the event of one or more host failures?

Requirements

1. All Clusters have a minimum requirement of N+1 redundancy
2. In the event of a host failure, a minimum level of performance must be guaranteed

Assumptions

1. vSphere 5.0 or later (Note: This is Significant as default reservation dropped from 256Mhz to 32Mhz, RAM remained at 0MB + overhead)

2. Percentage of Cluster resources reserved for HA is used and set to a value as per Example Architectural Decision – High Availability Admission Control

3. Strict admission control is enabled

4. Target over commitment Ratios are <=4:1 vCPU / Physical Cores | <=1.5 : 1 vRAM / Physical RAM

5. Physical CPU Core speed is >=2.0Ghz

6. Virtual machines sizes in the cluster will vary

7. A limited number of mission critical virtual machines may be set with reservations

8. Average VM size uses >2GB RAM

9. Clusters compute resources will be utilized at >=50%

Constraints

1. Ensuring all compute requirements are provided to Virtual machines during BAU

Motivation

1. Meet/Exceed availability requirements
2. Minimize complexity
3. Ensure the target availability and performance is maintained without significantly compromising  over commitment ratios

Architectural Decision

Ensure all clusters remain configured with the HA admission control setting use
“Enable – Do not power on virtual machines that violate availability constraints”

and

Use “Percentage of Cluster resources reserved for HA” for the admission control policy with the percentage value based on the following Architectural Decision – High Availability Admission Control

Configure the following HA Advanced Settings

1. “das.vmMemoryMinMB” with a value of “1024″
2. “das.vmCpuMinMHz” with a value of “512”

Justification

1. Enabling admission control is critical to ensure the required level of availability.
2. The “Percentage of cluster resources reserved for HA” setting allows a suitable percentage value of cluster resources to reserved depending on the size of each cluster to maintain N+1
3.The potentially inefficient slot size calculation used with “Host Failures cluster tolerates” does not suit clusters where virtual machines sizes vary and/or where some mission Critical VMs require reservations

  • 4.
  • Using advanced settings “das.vmCpuMinMHz” & “das.vmMemoryMinMB” allows a minimum level of performance (per VM) to be guaranteed in the event of one or more host failures
  • 5.
  • Advanced settings have been configured to ensure the target over commit ratios are still achieved while ensuring a minimum level of resources in a the event of a host failure
  • 6.
  • Maintains an acceptable minimum level of performance in the event of a host failure without requiring the administrative overhead of setting and maintaining “reservations” at the Virtual machine level
  • 7.
  • Where no reservations are used, and advanced settings not configured, the default reservation would be 32Mhz and 0MB+ memory overhead is used. This would likely result in degraded performance in the event a host failure occurs.

Alternatives

1. Use “Specify a fail over host” and have one or more hosts specified
2. “Host Failures cluster tolerates” and set it to appropriate value depending on hosts per cluster without using advanced settings
3.Use higher Percentage values
4. Use Higher / Lower values for “das.vmMemoryMinMB” and “das.vmCpuMinMHz”
5. Set Virtual machine level reservations on all VMs

Implications

1. The “das.vmCpuMinMHz” advanced setting applies on a per VM basis, not a per vCPU basis, so VMs with multiple vCPUs will still only be guarenteed 512Mhz in a HA event

2. This will reduce the number of virtual machines that can be powered on within the cluster (in order to enforce the HA requirements)

CloudXClogo

 

 

Example Architectural Decision – Distributed Power Management (DPM) for Virtual Desktop Clusters

Problem Statement

In a VMware View (VDI) environment where the bulk of the workforce work between 8am and 6pm daily, how can vSphere be configured to minimize the power consumption without significant impact to the end user experience?

Assumptions

1. The bulk of the workforce work between 8am and 6pm daily
2. Most users login during a 2 hour window between 7:30 and 9:30 daily
3. Most users logoff during a 2 hour window between 4:30 and 6:30 daily
4. VMware View cluster maintains at least N+1 redundancy
5. VMware View cluster only runs desktop workloads
6. VMware View cluster size is >=5
7. VMware View cluster/s are configured with HA admission control policy of “Percentage of cluster resources reserved for HA” to avoid the potentially inefficient slot size calculation preventing hosts going into standby mode

Motivation

1. Reduce the power consumption
2. Align with Green IT strategies
3. Reduce the datacenter costs
4. Reduce the carbon footprint

Architectural Decision

Configure and enable DPM on all ESXi hosts with the power management set to “Automatic” and the DPM threshold set to “Apply priority 3 or higher recommendations” and set hosts 1,2 and 3 in the cluster not to enter standby mode.

Justification

1. As the bulk of the users are inactive outside of normal business hours, a significant power saving can be achieved
2. The users do not all login at once, which allows DPM to gradually start ESXi hosts (which were put into standby mode by DPM previously)
3. In the event the workload is unusually low on a given day, power savings can be realized without significant impact to the end user experience
4. Where a large number of users login unexpectedly early one morning, the impact to users will be minimal
5. DPM is configured to ensure a minimum of three (3)  ESXi hosts remain on at all times. This number is expected to be able to support all desktops within the environment under low load (ie: 80% of desktops at idle). This number can be adjusted if required.

Implications

1. In the unlikely event a large number of users logon unexpectedly early one morning, the impact to users may be experienced for the time it takes for one or more ESXi hosts to exit maintenance mode. This is generally <10mins for most servers.
2. Out of band interfaces such as DRAC / iLO / RSA or IMM interfaces (depending on host hardware type) will need to be configured and be accessible to vCenter and the ESXi hosts to enable DPM to function
3. As the “Percentage of cluster resources reserved for HA” setting is static (not dynamically adjusted by DPM) in the event of a host failure while one or more hosts are in standby mode, in unlikely event a VM attempts to power on before a host has been able to successful exit standby mode, the VM may fail to power on.
4. Where large percentages of Memory reservations are used (see Example AD – Memory Reservation for VDI) then ability for the for DPM to put one or more hosts into standby will be reduced. Where DPM is expected to be used, no more than 50% memory reservation should be configured to ensure maximum  memory overcommitment can be achieved without placing a significant overhead on the shared storage for vSwap files
5. Monitoring solutions may need to be customized/modified not to trigger an alarm for a host that is put into standby mode

Alternatives

1. Set a lower number of hosts to remain on to maximize power savings – This may result in higher impact to users first thing in the morning in the event of high concurrent logins
2. Set a higher number of host to remain on, however this will minimize power savings and give less value to the added complexity of setting up DPM (and associated out of band management interfaces)
3. Set the DPM threshold more aggressive to maximize power savings – This would likely result in some impact to VMs due to increased physical cores being available to the CPU scheduler and physical memory being available for VMs which may result in swapping