1 of 12

QUICK ESTIMATION OF IT INFRASTRUCTURE FOR VDI

balagur.blogspot.com

2 of 12

DISCLAIMER

2

This is not a professional or company approved sizer. It was made just for quick estimation of compute resources required for VDI infrastructure.

It can be used if there is a lack of input information or if the customer or salesperson ask to give a quick response.

Formulas and parameters are not locked. So, you can change and tune sizer as you need

The filled file can be provided to the customer as explanation of your choice.

3 of 12

STEPS OF SIZING

  • 3

Size

Size primary and backup storage independently

Get

Get number of nodes , number of ports on switches per site

Choose

Choose node parameters

Add

Add user VM parameters to “Server Nodes” worksheet

4 of 12

CHOOSING NODE PARAMETERS (OVERVIEW)

Worksheets “Server Nodes”

4

  • Here you choose
    • User VM parameters
    • Resource load limits.
    • Server node configuration
    • Number of HA nodes (usually one is a good choice)
    • Optionally power consumption and height for 1 node
  • As a result, you get
    • The number of nodes required for your infrastructure
    • The number of ports that required on switches to connect nodes
    • Optionally summarized power consumption and height

​

5 of 12

CHOOSING NODE PARAMETERS. STEP 1

User VM characteristics

  • 5
  • This sizer suppose that all user VMs have identical parameters
  • Choose for every VM
    • Number of virtual cores
    • Core oversubscription (6 is a good value)
    • Memory amount for every VM
    • Disk space available for every VM

6 of 12

CHOOSING NODE PARAMETERS. STEP 2

Load limits, number of sites, number of users

  • 6
  • Here you can choose, what is the load of nodes (CPU and Memory). Default is 70%. It’s not a good idea to load the all nodes on 100%. You need some reserve for unexpected workload. You can change value.
  • Choose the number of sites if you develop a distributed VDI infrastructure
  • Choose total number of user on all sites

​

​

7 of 12

CHOOSING NODE PARAMETERS. STEP 3

Node parameters

  • 7
  • Choose the parameters of future node
    • Number of cores in node
    • Memory amount
    • I/O ports
  • When you change node parameters, look at “Number of Servers” column. In row “Number of New Nodes Required” you can see, how many nodes are required. Sizer estimates number of nodes based on Cores and Memory requirements and choose maximum of two values.
  • Numbers of nodes calculated for Cores and Memory should be close to each other. If values are too different than it means that configuration is unbalanced.
  • Keep in mind that sizer calculate the number of node for every site, not total amount for all sites.
  • Sizer estimate that in case of disaster of one of the sites other alive sites should support all users
  • Number of HA nodes (usually one is a good choice)
  • Parameter “CPU Processor Model” is just for your reference

8 of 12

CHOOSING NODE PARAMETERS. STEP 4

Results and power values

  • 8
  • The results are:
    • number of nodes required on every site (“Total Number of new Nodes per site required” parameter)
    • total number of I/O ports on every site (“Server Network Port in Solutions” table)
  • Optionally height and power consumption cells (use Power Advisor for HPE Servers) can be filled.
  • These values are for every node. Number of nodes will be taken from “Total Number of new Nodes required” cell

9 of 12

STORAGE REQUIREMENTS

Worksheet “Storage”

  • 9
  • On this worksheet you can get requirements for primary and backup storage systems
  • Sizer expect that all VMs are persistent. So every VM takes all presented space. Usually most of VDI VMs are non-persistent. There are algorithms that allows to estimate compression when golden volumes and replicas are used. I do not implement it in this sizer yet. Instead of it “Dedup ratio / non-persisten VM compression ratio” parameter can be used.
  • VDI infrastructure require some extra space for service VMs. Usually, it is 3-7 TB. Put this value in “Required Capacity for infrastructure VMs, TB” field
  • “Capacity max Usage limit” parameter is the same as core and memory limits for the servers. It is not expected that storage system will be filled in 100%. 70% of load is a good value
  • A dirty estimation for StoreOnce backup storage is 1,5 of primary storage capacity. It should be enough for 1 year of backup. It can be changed if you need. I recommend to use Ninja Start wizards to size StoreOnce. If it is not possible use estimation for this sizer.
  • Do not forget to fill cells with port quantities for primary and backup arrays.
  • Usually, every controller on storage has its own management port. So, for 2-controller model you need 2 management ports, for 4-controller model you need 4 management ports
  • StoreOnce requires 2 management ports (iLO and system management)
  • Optionally technical characteristics can be filled. Take values from storage sizers

10 of 12

NETWORK REQUIREMENTS

Worksheet “Networking”

  • 10
  • On this worksheet you can find requirements for network switches (LAN and SAN)
  • If you fill port quantities for servers and storage correctly than “Total number of ports to interconnect solution” row shows the total number of SAN and Ethernet ports, required for solution
  • Remember that this is the total number of ports. Usually, pair of switches are used in solutions. So, every switch should have at least half number of ports calculated here
  • I expect that SAN and production Ethernet switches have management ports. So, they must be considered as well.
  • Optionally technical characteristics can be filled

​

11 of 12

TECHNICAL CHARACTERISTICS

Worksheet “Tech Characteristics”

  • 11
  • On this page you can find the summarized values of solution technical characteristics:
    • Total height in rack
    • Total power consumption
    • Total heat dissipation

12 of 12

THANK YOU

balagur.blogspot.com

12

Please send me a message if you find a bug or have idea, how to tune the systems.