Cost optimization
Citrix Director now introduces a new feature called Cost optimization, which helps you to analyze the usage of virtual machines and sessions effectively. This feature provides insightful visual representations on how to optimize the cost. It also helps you to eliminate unnecessary machines which help to reduce costs.
The Cost optimization page includes the following features:
Cost savings [Preview]
The Cost savings page provides a visual representation of infrastructure savings accrued over a selected period and forecasts expected savings for the remaining days. By analyzing machine usage and sessions, this page helps you to identify achieved savings and opportunities for cost reduction. This page offers:
- Insight into the optimization of infrastructure costs
- The amount saved
- Information on a range of scenarios that might result in exceeding projected costs
- Potential opportunities for identification and strategic planning to realize infrastructure cost savings
The machine usage typically increases during peak hours, demanding increased resources. The Autoscale feature helps you to reduce cost by automatically powering off unused machines during the low demand period. The projected savings depend on this usage pattern. However, costs might exceed expectations under specific circumstances, such as when unused machines are present or when there are failures in Autoscale power actions. Identifying the reasons behind these occurrences was previously challenging.
The Cost savings page provides you detailed information on different scenarios that might cause you to exceed the expected cost. Similarly, for example, if there are machines with no sessions, it shows you further opportunities for cost savings.
Click the Cost optimization tab from the left menu on the Home page. And then, click the Cost saving tab to access the Cost savings page.
The Cost savings page includes Estimated savings and Autoscale savings report.
Estimated savings: This graphical representation helps to evaluate the efficient utilization of infrastructure resources. The cost savings are displayed in the currency of the hypervisor or as a percentage of the cost incurred. You can view the results for the last:
- Seven days
- Thirty days
- Three months
- Six months
- Twelve months
The Estimated savings graph displays the following:
- Estimated savings - Displays the amount of savings achieved in infrastructure for the selected duration
- Power-managed machines - Displays the total number of power-managed machines
- Projected savings - Displays how much infrastructure savings can be done for the remaining duration
- Average logon duration - Displays the average time for a user to log on over the selected time duration.
Autoscale Savings report: This graphical representation provides the savings report for the delivery groups for which the Autoscale feature is configured and enabled. This report is applicable only for the power-managed machines. You can select a delivery group and then select data for:
- Today
- Last day
- Last week
- Next week
- This month
- Next month.
You can hover over any graphs to retrieve specific chart values.
Note:
If the delivery group is deleted, you can’t get any data in the Autoscale savings report.
You can drill down to the following options from the delivery groups drop-down list :
- All: Infrastructure details display all the delivery group information.
- Single-sessions: Infrastructure details for all delivery groups which are single-session machines along with the graphical representation. The graphical representation includes the machine power state, successful Autoscale power actions, and sessions when you hover over the chart.
- Multi-sessions: Infrastructure details for all delivery groups which are multi-session machines along with the graphical representation.
The data displays for the selected duration. The graphical representation of the Estimated savings is the actual savings of the selection made from the drop-down menu. When you hover over the peak points, the information about the following are displayed:
- Machine power state
- Successful Autoscale power actions
- Sessions
Based on the delivery group selected, you can view the following details:
-
Infrastructure details
- Total machines and Power managed machines - Displays the total number of machines and out of that how many machines are power-managed. This data provides you a further opportunity to consider converting the remaining non-power managed machines to power-managed machines, thus enabling the Autoscale feature for a more comprehensive coverage. The data in the Infrastructure details section shows the last known information updated every hour.
-
Catalog details
Displays the name of the hypervisors (service providers). For more information, see Supported VDA hosting platforms. The data in the Catalog details shows the most recent data (live data) available at the current time.
You can get the following details:
- Hypervisor name - Displays the name of the hypervisor to which the site is connected. Displays all available hypervisors when multiple hypervisors are present.
- Instance - Displays the name of the instance.
- Cost per hour - Displays the cost per machine per hour in the currency that the hypervisor supports. For Azure machines, the Azure retail cost is displayed. For more information, see Azure Retail Prices overview. For other machines, cost is displayed based on the cost entered by admins in the Web Studio. The default cost is $1 if the cost is not mentioned in Web Studio.
- Machines - Displays the count of the machines for the selected instance.
-
Autoscale configuration (last known)
Tag in use - Displays the tag of the machines that is configured in the Autoscale settings in a WebStudio. You can also filter the data based on the tagged machines. In this case, you get only the tagged machines summary.
Capacity buffer (peak) and Capacity buffer (off peak)
Displays the percentage of buffer for machines that are powered on based on the maximum requirement of the delivery group during peak and off-peak times.
-
Estimated savings
Displays the amount of savings achieved in infrastructure for the selected duration.
-
Power managed machines
With Autoscale enabled, the power-managed machines are turned off when not in use to optimize costs. You can get the following details:
- Number of machines powered off - This data helps you to analyze the number of machines that were powered off and thus helped to reduce cost.
- Hibernate state - This data helps you to analyze how many machines are in the hibernated state.
- Machines on with no sessions - This data provides you the opportunity to analyze why machines were active without any sessions. This data provides you the opportunity to plan for a more effective infrastructure savings strategy.
- Machines on with sessions - This data helps you to analyze the number of machines that were powered on and with active sessions.
- Power off - This data helps you to analyze the number of machines that were powered off or suspended and when this action was done. You can verify this data and make further modifications, contributing cost reduction.
- Power on - This data helps you to analyze how many machines were powered on and when this action was done. You can verify this data and make further modifications and thus help to reduce cost.
-
Concurrent sessions
Displays the total number of sessions at the selected time. This data helps you to determine when to power off machines for cost optimization.
-
Average logon duration
Displays the average time for a user to log on over the selected time duration.
Infrastructure rightsizing
The Infrastructure Rightsizing page helps you to analyze the provisioning and sizing aspects of your delivery group based on resource utilization. Based on this analysis, you can:
- Reduce spending on the unused resources
- Identify if the machines for the given delivery group consumption are over, under, or right sized
- Identify if the utilization is mainly driven by Power users (indicated by higher utilization by top 5 percentile machines compared to the rest of the group)
You can filter the Infrastructure Rightsizing using the following:
- Delivery Group - You can filter for single-session OS or multi-session OS delivery groups
- Tags - Tags are the tag names applied to the machine. So, you can filter machines with the same tags. You can select multiple tags up to five tags. When you select multiple tags, you can filter for all the machines which have at least one of those selected machine tags applied.
- Time period - You can filter the data for the last 24 hours, 7 days, and 30 days.
The Infrastructure rightsizing page offers:
- Insight in to the utilization details
- Summary of the resource utilization
- Trends on the resource utilization
The aggregated data displayed on this page is updated as follows:
- One hour for Last 24 hours, and Last 7 days filter
- One day for Last 30 days filter
Click the Cost optimization tab from the left menu on the Home page. And then, click the Infrastructure rightsizing tab to access the Infrastructure rightsizing page.
You can also click the Rightsize this delivery group link from the Infrastructure details section in the Cost optimization tab > Cost savings to access the Infrastructure rightsizing page.
The Infrastructure rightsizing page includes:
- Utilization details
- Resource utilization summary
- Resource utilization trends
Utilization details: This section provides details on the utilization of the selected delivery group and for the selected tags. You can view the details for the last 24 hours, 7 days, and 30 days. The Utilization details section provides details for the entire duration of the selected time filter. For example, if 7 days is selected, the total machine count is the total number of machines that were on at any point during the 7 days. The Utilization details section provides the following:
- Total machine - Displays the total number of machines. This count includes all the machines which were powered on during the selected period. For example, If 7 days is selected, the total machine count is the total number of machines that were on at any point during the 7 days.
- Average session density - This graph is an average across the number of machines and across the selected time period. Refers to the amount of session load on a machine which affects the resources available for sessions.
- Average machine uptime - It’s the ratio of total uptime of the power managed machines and for the selected time period. How much time the machine was up, per day per machine. Suppose that there are two machines in one delivery group or in the selected tag, and one machine was used for 24 hours and the other machine was used for 1 hour, the average machine uptime is 12 hours 30 minutes. In this example, the selected filter value is Last 24 hours.
- Unused machines - Displays the number of machines which had no sessions over the selected time period but which was still powered on. Click the Unused Machines link to open the Filters tab where you can view all the machines in that same delivery group which are currently powered on and have no sessions.
Resource utilization summary: This section provides a summary on how the resources were used for the selected delivery group and for the selected tags. You can select multiple tags up to five tags. You can view the details for the last 24 hours, 7 days, and 30 days. The machines are ranked based on the utilization. The Resource utilization summary section provides details on CPU utilization and memory utilization for the machines with active sessions.
-
CPU utilization:
- Machine cohort - Provides information about the machine groups categorized into top, middle, and bottom based on its CPU utilization. 5% of the most used machines are categorized under Top 5 percentile, 5% of the least used machines are categorized under Bottom 5 percentile, and the remaining 90% are categorized under Middle 90 percentile.
- Range of CPU Peaks - Displays the CPU peaks range based on its utilization corresponding to the machine cohort.
- Machines - Displays the number of machines corresponding to the machine cohort.
- Session density - Displays the session density corresponding to the machine cohort.
-
Memory utilization:
- Machine cohort - Provides information about the machine groups categorized into top, middle, and bottom based on its memory utilization. The 5% of the most used machines are categorized under Top 5 percentile, 5% of the least used machines are categorized under Bottom 5 percentile, and the remaining 90% are categorized under Middle 90 percentile.
- Range of memory Peaks - Displays the memory peaks range based on its utilization corresponding to the machine cohort.
- Machines - Displays the number of machines corresponding to the machine cohort.
- Session density - Displays the session density corresponding to the machine cohort.
The following table provides information on the possible action items that you can do based on the values present in the Resource Utilization summary section:
Scenario | Action item |
---|---|
CPU utilization is more | Increase the CPU specifications of the delivery group. Or, identify the higher usage users and move them to a different delivery group with better specifications. |
CPU utilization is less | Decrease the CPU specification of the delivery group. |
Memory utilization is more | Increase the memory specifications of the delivery group. |
Memory utilization is less | Decrease the memory specification of the delivery group. |
How is the data populated in the Resource utilization summary section?
The data in the 8*Resource utilization summary section is populated based on the data in the Resource utilization trends. The summary gives you details only for the machines with active sessions. The summary gives you better results when there are more than five machines.
Based on the number of machines in the selected delivery group, the maximum and minimum utilization value is calculated.
For example, consider that you have selected the Last 7 days as a time period and you have 20 machines in the selected delivery group. These 20 machines are further categorized into Top 5 percentile, Middle 90 percentile, and Bottom 5 percentile. The CPU and memory utilization is calculated when there are active sessions. For example, in this case, the first and last machines are categorized into the top 5 and bottom 5. The remaining 18 machines are categorized into the middle 90 percentile. In this case since there is only one machine in the Top 5 percentile, the maximum and minimum value is the same. Consider that there is more than one machine in the group. In this case, the highest and lowest value of the group is based on the maximum value of the highest-ranked machine in that group, and the maximum value of the lowest-ranked machine in that group respectively.
When you select CPU as the Resource type, the Range of CPU peaks and the Peak memory utilization for the selected machines are displayed.
When you select Memory as the Resource type, the Range of memory peaks and the Peak CPU utilization for the selected machines are displayed.
Resource utilization trends: This section provides a graphical representation for the Range of CPU peaks, peak memory utilization, average Session density, average ICA RTT, and average logon duration.
You can filter the details for the following:
- Resource - CPU and Memory
- Machine cohorts - Top 5 percentile, Middle 90 percentile, and Bottom 5 percentile
- Trends - Average session, Average ICA RTT, Average logon duration.
You can view the following charts:
- Range of CPU peaks - Displays the range of CPU peaks in a band. When you hover on any point in the band, the top and bottom value of the band is displayed.
- Peak memory utilization - Displays the highest memory utilization for the selected machines and for the selected time over a graph.
- Average Session density - Displays the average session density for the selected machines and for the selected time over a graph.
- Average ICA RTT - Displays the average round trip-time for the session. This chart shows breaks when there are no active sessions or when sessions are disconnected.
- Average logon duration - Displays the average logon duration for the selected machines and for the selected time period over a graph. This chart shows data when a new logon is started during the selected time.