ishenwei/nexus
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
b40abbcd473a7093d8261e212e3d6de97c1e516a
nexus/knowledgebase/csd-wiki/ICSD/ESM-Cloud-Infra-Cost-Review_686065545.md

18 KiB
Raw Blame History

ESM-Cloud-Infra-Cost-Review_686065545

Introduction

This page presents all the ESM SaaS related cost review results.

2025-07-15

eu8+18: 18 unsued ebs reduce 250$ monthly +140 ebs's on gp2 which need to go on gp3=250$monthly REVIEWING

2025-05-13

  1. [EU-Managed eu28-eu32] Change EBS from gp2 to gp3, save $64.00 per month REVIEWING
  2. [EU-Managed eu28-eu32] Remove unused EBS, save $10 per month REVIEWING
  3. [EU-Managed eu28-eu32] Upgrade the EKS version to 1.32 before 2025-07-23 for eu28 and eu32, to prevent $1000/mo extra cost on extend support REVIEWING Owner: Maricel EU28 upgrade planned on 2002-07-20, EU32 upgrade planned on 2025-07-17

4. [EU-Managed eu28-eu32] Remove SageMaker, save $240/mo REVIEWING

2025-05-07

  1. Remove SageMaker, save REVIEWING Owner: Ting

2025-04-07

  1. Set remove policy for EBS snapshots, save around $100/mo, the problem is that this cost is keep growing. NO PLAN Owner: Ting Checked that it's managed by velero.
    example: us24-prod-eks-cluster-dynamic-pvc-29b4c73e-edb7-49e1-9ca9-ce653f04ed57,
  2. Upgrade the EKS version to 1.32 before 2025-07-23, to prevent $6000/mo extra cost on extend support REVIEWING Owner: Ting, change together with CCoE new image, after 25.2 upgrade.
  3. US2-DEV
    1. Change the op RDS from t3.xlarge > t4g.large, as the cpu usage is less than 5%, save around $100/mo depends on the usage REVIEWING Owner: Ting ask Sunny for a change. 2. Change the op RDS disk from io2 to gp3, as gp3 is cheaper and provide more IOPS, save around $ 500 /mo REVIEWING Owner: Ting
      1. make sure the future op database start with t4g.large and gp3 2. may be we can remove few instances, as there are 3 OP related RDS. NO PLAN
      2. turn off the RDS if the system is not in use. save around $100~500/mo NO PLAN
      3. remove manual RDS snapshot keep recent 30 days: $300+ REVIEWING Owner: Ting
  4. Apply the ecr storage cleanup solution: $ 500DONE owner: Ting Ye

2025-03-28

Reorganize the FinOps related tasks:

  1. EU18
    1. Migrate "Tech Mahindra - YIT Prod-1947697" to new FinOps and then remove the old vertica, save around $1300/m REVIEWING owner: shenwei to check with PM
  2. US24
    1. Remove the old vertica, save around $ 1500 /m It's stopped for a month, owner: Ting DONE 11 Apr 2025
  3. EU3
    1. Migrate 4 remaining tenants from FinOps_Classic to new FinOps and then remove the old vertica, save around $1300/mo REVIEWING owner: shenwei to check with presales
  4. US2
    1. Decommission 920775298 and 263660258 which are using FinOps_Classic, then remove the vertica, save around $ 500 /mo REVIEWING Lingyan has confirmed with Alex Dominic Savio William. Owner: shenwei

2025-03-25

  1. Add RI and SP on EU28, saved around $1300/mo DONE 25 Mar 2025
  2. Add RI for OpenSearch on US2-PROD (Covering US2/US6), save 339$/mo DONE 25 Mar 2025
  3. Add RI for OpenSearch on EU8, save 339$/mo. DONE 25 Mar 2025
  4. Based on the system usage, resize us2 smax RDS m6g.4xlarge → r6g.2xlarge, cms RDS r6g.2xlarge → r6g.xlarge, save around $ 1100 /mo DONE

2025-02-27

  1. Change all the EC2 instances from m5/r5 to m6i/r6i with better performance REVIEWING
    1. Asked RnD to test the new instance type. if we choose m6a/r6a can save around $3000+/m without performance improvements Feature 2481012 - [Cost saving] [SaaS] Suppor AMD EC2 servers which can save up to $3000/month on SaaS 2. And m6g/r6g can save even more like 9000$, but most likely it's not working, pending RnD for testing. NO PLAN Danny: Arm-Based EC2 cannot be used for worker node, because our applications are not built for ARM processors!

2025-01-08

  1. Cleanup SMAX EFS, save around $ 1500/m, owner: Ting Ye
    JP12-STG,EU3,US7,US2-Prod,US6,EU8,AP10 DONE 28 Mar 2025
  2. US6
    1. After decommission Ford, resize the farm, save around $400/mo DONE
      Also Dick's Sporting good just onboard, we may not change the sizing, but make sure to remove files on efs.
  3. EU3
    1. Decommission legacy Carbon server and ALBs, save around $200/m DONE

2024-12-11

  1. EU18
    1. Change the op RDS from t3.xlarge > t3.large, as the cpu usage is less than 5%, save around $120/m DONE 2. Change the RDS disk from io2 to gp3, as gp3 is cheaper and provide more IOPS, save around $350/m DONE
  2. EU3
    1. Change the op RDS from t3.xlarge > t3.large, as the cpu usage is less than 5%, save around $120/m DONE 2. Change the RDS disk from io2 to gp3, as gp3 is cheaper and provide more IOPS, save around $350/m DONE
  3. us24
    1. Change the op RDS from t3.xlarge > t3.large, as the cpu usage is less than 5%, save around $120/m DONE
  4. RI for OP RDS save around $100/m DONE

2024-09-20

  1. Terminate the old bastion node. Save around $300/m. DONE

2024-08-28

  1. Turn off the auto RDS backup, as we have the "AWS Backup service". Save around $300/m. DONE
    1. us2-prod oomt, eu18-prod oo/cms/smax, ap10 oomt, ca16 oo/cms/smax, jp12-stg cms/oo/smax.
  2. Finish the Helm Post-transformation tasks, save more than $100/m. REVIEWING
  3. Remove the OMT ingress (16.43$ * 12 = $197). Reviewing the situation with RnD, RnD may provide a step to do it. REVIEWING
    https://docs.microfocus.com/doc/SMAX/24.2/TransformSmaxToHelmBased#Clean_up_OMT_resources_in_the_OMT_namespace Issue 2323030 - [Doc] [SaaS] Remove unnessary OMT resoruces after helm transformation.

2024-07-08

  1. Clean up the RDS backup tables in the database (not saving money but preventing the cost to increase) REVIEWING
    1. Audit (rnd review) 2. revinfo (rnd confirmed to truncate) need a change 3. bak tables (ops review)

2024-05-14

  1. Change all workers volume type from gp2 to gp3. Save around $ 1000/m. Plan with CCoE ami change PLANNED
  2. OpenSearch on US2-Dev Around 800$ Plan to decommission DONE owner: Ting
  3. EU18
    1. Reduce the vertica data node size from m4.4xlarge to r5.2xlarge: Save $600/m NO PLAN owner: Scott
  4. Review the backup procedure for potential issues on cost
    1. Change the kms key os RDS, so that we can enable incremental backup. Can save around $ 2000 +/m REVIEWING Solution is ready, requires a downtime
  5. Need to check the saving plan usage 80%~90%, need to check with Vinay. List the number DONE 25 Mar 2025 by lingyan, coverage 99% by the company
  6. Remove the us2-dev EKS smax-cluster-us2dev in 551360491749: $300/m DONE By Ting
  7. Review the saving plan for us26 Owner:Lingyan DONE 25 Mar 2025 by lingyan, coverage 99% by the company

2024-05-08

  1. Change cross-region retention from 14 days to 7 days DONE Yu Liu
  2. Remove the local backup generated by cross-region backup owner: DONE Yu Liu
  3. Cleanup SMAX EFS to reduce the cost: $ 2000, owner: Ting Ye
    ,JP12-STG,EU3,US7,US2-Prod,US6,EU8,AP10 DONE
  4. Checking if we can change the backup type to cold backup for efs owner: NO PLAN Yu Liu
  5. Remove the us1 EKS in 361684190412, as it cost $400 for eks extended support owner: Ting Ye DONE
  6. check the ecr storage cleanup solution: $ 800 owner: Ting Ye
    --US2-dev,EU3-Prod,US7-Prod DONE
  7. Check the CMS efs cleanup solution with rnd owner: Ling-yan Meng DONE Clean up CMS log files
    1. Cleanup CMS EFS to reduce the cost: $ 500 owner: Ting Ye DONE 2. US2-dev,US2-Prod,EU3,US7,US24 Done 3. JP12,BR14,CA16,US26 Done

2024-04-28

  1. US6/EU8/AP10/EU18
    1. EU8
      1. remove not used tenant-export/tenant-import 400GB packages: 200$ DONE

2024-03-25

  1. US6/EU8/AP10/EU18
    1. AP10
      1. remove manual RDS snapshot keep recent 30 days: $85 DONE 2. remove unused ALB (save 16.43$ * 7 = 115$ ) Check FQDN and traffic, peer review before removal REVIEWING
        1. acd2b58c6b3fc40a3a911c79dd0f8105-7bb8644d3ea49e82.elb.ap-southeast-2.amazonaws.com (k8s-itsmatbx-itomngin-1af9765940) 2. internal-SMAX-EKS-ALB-832470617.ap-southeast-2.elb.amazonaws.com 3. internal-CMS-ALB-1420616780.ap-southeast-2.elb.amazonaws.com 4. internal-cms-smax-integration-605090270.ap-southeast-2.elb.amazonaws.com (should be the legacy integration) 5. internal-k8s-ap10prodcmsalb-99eed8dbd4-1596301489.ap-southeast-2.elb.amazonaws.com 6. internal-k8s-oopublic-e2b012afab-1414584707.ap-southeast-2.elb.amazonaws.com 7. internal-k8s-ap10auditalb-f2e1f6a5de-476511256.ap-southeast-2.elb.amazonaws.com
      2. US6
      3. remove manual RDS snapshot keep recent 30 days: $ 300 + DONE 2. remove unused ALB (save 16.43$ * 8 = 131$ ) Check FQDN and traffic, peer review before removal REVIEWING
        1. ad2b5ab2128d842a4ab7a8479b91d6ca-5f7aeea4304fc4ab.elb.us-west-2.amazonaws.com (k8s-itsmaohs-itomngin-b928fd8ff6) 2. internal-SMAX-ALB-1780068998.us-west-2.elb.amazonaws.com 3. internal-subdomain-testing-1383237556.us-west-2.elb.amazonaws.com 4. internal-CMS-ALB-103193064.us-west-2.elb.amazonaws.com 5. internal-ALB-For-Integration-1506362286.us-west-2.elb.amazonaws.com 6. internal-k8s-us6prodcmsalb-05d13e29f6-1782663167.us-west-2.elb.amazonaws.com 7. internal-k8s-oomtpublic-8b587340e7-989485052.us-west-2.elb.amazonaws.com 8. internal-k8s-us6auditalb-b4c1ac47bd-1257080049.us-west-2.elb.amazonaws.com
      4. EU8
      5. remove manual RDS snapshot keep recent 30 days: $ 400 + DONE 2. remove unused ALB (save 16.43$ * 8 = 131$ ) Check FQDN and traffic, peer review before removal REVIEWING
        1. internal-CMS-ALB-EU8-50066461.eu-central-1.elb.amazonaws.com 2. internal-k8s-eu8cmsext-09c603805a-1635150560.eu-central-1.elb.amazonaws.com 3. internal-EU8-ALB-For-Integration-1099466715.eu-central-1.elb.amazonaws.com 4. af67eb4c5555d47aab1230aaeafbfcfd-82a881994f2d96b7.elb.eu-central-1.amazonaws.com (k8s-itsmah3c-itomngin-c4e78faf0f) 5. internal-SMAX-ALB-582960966.eu-central-1.elb.amazonaws.com 6. internal-EU8-ALB-For-Integration-1099466715.eu-central-1.elb.amazonaws.com 7. aff85e03390924d0c9a6eae56cf2b525-6a24d3a3cd2ad390.elb.eu-central-1.amazonaws.com (this one has traffic on 80, k8s-itsmah3c-itomngin-21d82011c2) 8. internal-k8s-oomtpublic-8f26407304-488986183.eu-central-1.elb.amazonaws.com
      6. EU18
      7. US6-STG
      8. (us-east-1) remove manual RDS snapshot keep recent 30 days: $40+ DONE 2. (us-west-2) remove manual RDS snapshot keep recent 30 days: $100+ DONE
  2. JP12/BR14/CA16
    1. JP12
      1. remove manual RDS snapshot keep recent 30 days: $70 DONE
      2. BR14
      3. CA16
      4. jp12-stg
      5. remove manual RDS snapshot keep recent 30 days: $40 DONE
  3. US2/US2-DEV/US24
    1. US2
      1. remove manual RDS snapshot keep recent 30 days: $ 1000 + DONE
      2. US2-DEV
      3. remove manual RDS snapshot keep recent 30 days: $200+ DONE 2. (us-east-1) remove manual RDS snapshot keep recent 30 days: $20+ DONE
      4. US24
      5. remove manual RDS snapshot keep recent 30 days: $20 DONE
  4. EU3/US7
    1. EU3
      1. remove manual RDS snapshot keep recent 30 days: $ 500 + DONE
      2. US7
      3. remove manual RDS snapshot keep recent 30 days: $ 500 + DONE
  5. EU22/US26
    1. EU22 2. US26
      1. remove manual RDS snapshot keep recent 30 days DONE 2. CMS RDS r6g.4xlarge -> r6g.2xlarge (save $800) last 4 weeks peak CPU 10% DONE

2023-12-19

Make sure to check and keep the max_connections

  1. US6/EU8/AP10/EU18
    1. AP10
      1. OO RDS: m6g.2xlarge → r6g.xlarge (save $150) last 4 weeks peak CPU 12% DONE
      2. US6
      3. OO RDS: m6g.2xlarge → r6g.xlarge (save $150) last 4 weeks peak CPU 6% DONE
      4. EU8
      5. OO RDS
        1. m6g.2xlarge -> r6g.xlarge (save $150) last 4 weeks peak CPU 15% DONE 2. IO1 3000 -> GP3 IOPS 12000 (save $300) DONE
      6. EU18
      7. SMAX RDS m6g.4xlarge -> r6g.2xlarge (save $300) last 4 weeks peak CPU 18% DONE 2. OO RDS: m6g.2xlarge → r6g.xlarge (save $150) last 4 weeks peak CPU 9% DONE
  2. JP12/BR14/CA16
    1. JP12
      1. SMAX RDS m6g.2xlarge -> r6g.xlarge (save $150) last 4 weeks peak CPU 13% DONE 2. CMS RDS m6g.2xlarge -> r6g.xlarge (save $150) last 4 weeks peak CPU 6% DONE 3. OO RDS: m6g.2xlarge → r6g.xlarge (save $150) last 4 weeks peak CPU 14% DONE
      2. BR14
      3. CMS RDS m6g.2xlarge -> r6g.xlarge (save $150) last 4 weeks peak CPU 9% DONE 2. OO RDS: m6g.2xlarge → r6g.xlarge (save $150) last 4 weeks peak CPU 5% DONE
      4. CA16
      5. SMAX RDS m6g.2xlarge -> r6g.xlarge (save $150) last 4 weeks peak CPU 5% DONE 2. OO RDS: m6g.2xlarge → r6g.xlarge (save $150) last 4 weeks peak CPU 4% DONE
      6. jp12-stg
      7. OO RDS: m5.2xlarge → r6g.xlarge DONE
  3. US2/US24
    1. US2
      1. OO RDS: m6g.2xlarge → r6g.xlarge (save $150) last 4 weeks peak CPU 19% DONE
  4. EU3/US7
    1. EU3
      1. CMS RDS m6g.2xlarge -> r6g.xlarge (save $150) last 4 weeks peak CPU 15% DONE 2. OO RDS: m6g.2xlarge → r6g.xlarge (save $150) last 4 weeks peak CPU 17% DONE
      2. US7
      3. OO RDS: m6g.2xlarge → r6g.xlarge (save $150) last 4 weeks peak CPU 10% DONE

2023-11-30

  1. US6/EU8/AP10/EU18
    1. US6
      1. CMS RDS: IO1 10000 IOPS → GP3 IOPS ($1093.5 → $86.25) DONE 2. SMA RDS:
        1. r6g.4xlarge → r6g.2xlarge (save $1312) Last 4 week free memory more than 80G, CPU less than 30%. Check and keep the max_connections DONE 2. GP2 → GP3 DONE 3. ~~CMS RDS: m6g.4xlarge → r6g.2xlarge ($823 -> $741 → $463) Need to wait till Dec 18th for RI expiration for $463 Last 4 weeks CPU < 30%, Min Free memory 31G
          ~~As the load on 7th Dec is higher than 60%. 4. Remove manual & backup snapshots, keep recent 30 days: $400+ DONE
      2. EU8
      3. SMA RDS: GP2 4500G → GP3 4500G 18000 IOPS 1000 MBPS ($1229 → $1616 ) Last 4 week IOPS peak time: 17000-18000, MBPS peak time: 450 MBPS DONE As the us6 worked, we can plan the eu8 change
        1. We can only switch to 18000 IOPS 500 MBPS at first and keep monitoring, if required, increase to 1000 MBPS, it will require less than $100 per month. Based on the monitoring, we need to change the MBPS from 500 MBPS to 1000 MBPS. DONE 2. Remove manual & backup snapshots keep recent 30 days: $1000+DONE 3. Disable EFS throughput mode for monitoring: $350 DONE
      4. AP10
      5. Remove manual & backup snapshots keep recent 30 days: $100+ DONE
  2. JP12/BR14/CA16
    1. JP12
      1. Remove manual & backup snapshots keep recent 30 days: $50+DONE
      2. BR14
      3. Remove manual & backup snapshots keep recent 30 days: $150+DONE
      4. CA16
      5. Remove manual & backup snapshots keep recent 30 days: $50+DONE
  3. US2/US24
    1. US2
      1. SMA RDS: IO1 2000G 3000 IOPS → GP3 2000G 12000 IOPS ($1100→ $460) Last 4 week IOPS peak time: 2500-4000, MBPS peak time: 70-150 MBPS DONE 2. CMS RDS: IO1 500G 3000 IOPS → GP3 500G 12000 IOPS ($1100→ $460) Last 4 week IOPS peak time: 2500-4000, MBPS peak time: 70-150 MBPSDONE
      2. US24
      3. SMA RDS:
        1. m6g.2xlarge → r6g.xlarge ($998→ $726) DONE 2. disable multi-AZ: save $363 DONE 2. OO RDS: m6g.2xlarge → r6g.xlarge ($499 →$363) Check and keep the max_connections DONE 3. Vertica
        2. Reduce the vertica data node number from 3 to 1 2. Reduce the vertica data node size from r5.8xlarge to r5.4xlarge: $1600 DONE

Need to request RI before Dec 18th. Wei Shen

Improve the instance type to newer version. RnD

Backup policy: keep only one month.

Check us2 tenant FinOps usage, may be it's been moved to us24?

https://us2-smax.saas.microfocus.com:443/saw/ess?TENANTID=920775298

Posted by lmeng2 at Mar 28, 2025 02:39 EDT

Reference in New Issue View Git Blame Copy Permalink