Optimize computing resource configuration for cost reduction and performance improvement - MaxCompute

The computing resource optimization feature analyzes your actual job requests and resource allocation expectations to generate an optimal resource configuration plan for level-1 subscription quotas. This topic walks through two common scenarios to show how the feature helps you cut computing costs or eliminate job delays.

For a detailed description of the feature, see Optimization of computing resource configuration.

Usage notes

Prices in this topic are for reference only. See the product page for current pricing.
After you receive a recommended plan, adjust your resource configuration gradually rather than switching all at once. This reduces the risk of instability during the transition.

When to use each scenario

The following table maps your current situation to the relevant scenario.

Situation	Expected outcome	Scenario
Reserved compute units (CUs) are idle and the monthly cost is high	Cut monthly cost by up to ~71.7%	Scenario: Idle resources with high cost
Reserved CUs are insufficient and important jobs miss their deadlines	Eliminate job delays with a minimal cost increase	Scenario: Insufficient resources with job delays

Scenario: Idle resources with high cost

A company in the early stage of its data warehouse build purchases 200 subscription reserved CUs to guarantee that 520 daily jobs finish before 08:00. The jobs complete on time — sometimes ahead of schedule — but the monthly computing cost reaches USD 4,400.

After a cost-reduction OKR (Objectives and Key Results) is issued, a data O&M engineer uses the computing resource optimization feature to find the right-sized plan.

Step 1: View current resource utilization

Log on to the MaxCompute console. In the top navigation bar, select a region. In the left-side navigation pane, choose Intelligent Optimization > Computing Resource Config Optimization.
On the Computing Resource Config Optimization page, select a level-1 subscription quota from the Quota Name drop-down list. The Estimated Daily CU Requests chart reflects the current job pattern: jobs run every hour, with a spike from 05:00 to 08:00.

Step 2: Set estimation time points

In the Set Estimation Time Point step, add 05:00 and 08:00 as estimation time points. This tells the system to simulate whether all jobs initiated before 05:00 finish by 05:00, and whether important jobs initiated from 05:00 to 08:00 finish by 08:00.

Step 3: Check the current plan

Click Current Plan Estimation to simulate job completion under the current configuration.

The CU Consumption Simulation (Current Plan Estimation) chart shows no delays at either time point, but significant idle capacity. This confirms that cost reduction is feasible.

Step 4: Generate a recommended plan

In the Set Optimization Goal section, the estimation time points are pre-filled with the same values. Click Generate Recommended Plan.

Step 5: Review the recommended plan

The CU Consumption Simulation (Recommended Plan) chart shows the optimized configuration:

50 reserved CUs (always on)
50 elastically reserved CUs from 04:00 to 05:00 and from 06:00 to 08:00

With this plan, all important jobs still finish before 08:00, and the monthly cost drops to USD 1,319.6 — a reduction of about 70%.

Step 6: Adjust the goal if needed

If a 30-minute delay on job output is acceptable, return to Set Optimization Goal and change the optimization goal for the 08:00 time point to 08:30, then click Generate Recommended Plan again.

The updated plan reduces elastically reserved CUs to the period from 06:00 to 07:00 instead of 06:00 to 08:00. Jobs finish before 08:30, and the monthly cost drops further to USD 1,246.4 — a reduction of about 71.7%.

Step 7: Apply the plan gradually

Important

Apply the recommended plan in stages rather than all at once. An abrupt reduction can cause instability if job volumes are higher than estimated.

The company first reduces reserved CUs from 200 to 100 and monitors job completion for a period. After confirming that job volumes have not increased significantly and that the system still recommends further reduction, the company applies the full recommended plan.

Step 8: Verify the results

After the trial period, confirm that the optimization achieved its goals:

Check that important jobs consistently finish before 08:00 (or 08:30 if you adjusted the goal).
Review the monthly cost invoice to confirm the reduction from USD 4,400 to approximately USD 1,246–1,320.
If job delays reappear, re-run Current Plan Estimation to check whether workloads have grown beyond the new plan's capacity.

Scenario: Insufficient resources with job delays

A company purchases 60 subscription CUs for 520 daily jobs. As the data scanned by jobs grows, important jobs initiated from 05:00 to 08:00 miss the 08:00 deadline. Jobs initiated before 05:00 also see a 3-minute delay. The company wants to fix the delays without substantially increasing costs.

Step 1: View current resource utilization

Log on to the MaxCompute console. In the top navigation bar, select a region. In the left-side navigation pane, choose Intelligent Optimization > Computing Resource Config Optimization.
On the Computing Resource Config Optimization page, select the level-1 subscription quota from the Quota Name drop-down list. The Estimated Daily CU Requests chart shows a spike in requests from 05:00 to 08:00.

Step 2: Set estimation time points

In the Set Estimation Time Point step, add 05:00 and 08:00 as estimation time points.

Step 3: Check the current plan

Click Current Plan Estimation to simulate job completion under the current 60-CU configuration.

The CU Consumption Simulation (Current Plan Estimation) chart confirms the delays:

Jobs before 05:00: 3-minute delay
Jobs from 05:00 to 08:00: 48-minute delay

This matches the delays observed in production.

Step 4: Generate a recommended plan

In the Set Optimization Goal section, the estimation time points are pre-filled with the same values. Click Generate Recommended Plan.

Step 5: Review the recommended plan

The CU Consumption Simulation (Recommended Plan) chart shows no delays at either time point under the optimized configuration:

50 reserved CUs (always on)
50 elastically reserved CUs from 04:00 to 05:00 and from 06:00 to 08:00

After the optimization, the monthly cost is USD 0.4 less than the current cost.

Step 6: Apply the recommended plan

Important

Apply the recommended plan in stages rather than all at once. An abrupt change in reserved CUs can cause instability if job volumes spike unexpectedly.

The company decides to add elastically reserved CUs as recommended but not reduce the number of reserved CUs yet, to minimize risk during the transition.

In the left-side navigation pane of the MaxCompute console, choose Workspace > Quotas.
On the Quotas page, find the level-1 quota and click Quota Configuration in the Actions column.
On the Quota Plans tab of the Quota Configuration page, click Add Plan.
In the Create Quota Plan dialog box, set Elastically Reserved CUs to 50 and click OK.

Configure the following scheduling plan based on the time points in the recommended plan. For more information, see Configure quotas.

Start time	Quota plan
00:00	Default
04:00	The quota plan created in the previous step
05:00	Default
06:00	The quota plan created in the previous step
08:00	Default

Note

The Default plan has Elastically Reserved CUs set to 0.

Step 7: Verify the results

After the trial period, confirm that the optimization achieved its goals:

Check that important jobs consistently finish before 08:00.
Confirm that the monthly cost has decreased by approximately USD 0.4 compared to the previous 60-CU configuration.
If delays reappear, re-run Current Plan Estimation to determine whether workloads have grown beyond the new plan's capacity and whether a further adjustment is needed.

What's next

To learn more about configuring quotas, see Configure quotas.
To understand how the computing resource optimization feature works, see Optimization of computing resource configuration.