Files
nexus/knowledgebase/csd-wiki/ICSD/ITOM-SaaS-Pain-Points_686083998.md

3.4 KiB
Raw Blame History

ITOM-SaaS-Pain-Points_686083998

Introduction

This page presents all the pain points for SaaS delivery on the ITOM BU level.

Legends

SMAX

HCMX

FINOPS

OO

OPS

Pain points

  • Process
    • OPS Dont wait until retro, we need a system that can record such inputs.
  • Quality
    • Recent system crashes caused by the new features. (Incident ID, Feature ID?)
      • SMAX SMAX platform pod loads are not well balanced. (24.2 may fix it)
      • It's not...
      • SMAX nativeSACM consumes high resource usage, especially network.
      • SMAX The SMAX still has OOTB index issues.
      • SMAX Redis is the single point of failure and is not easy to debug.
      • SMAX HCMX FINOPS OO Need more post upgrade tests. / CMS post upgrade issue is not acceptable
      • SMAX SLT task is weekly (unplanned change is not submitted )
      • After 2024.1, CMSs quality is not good. So many issues.
  • SLA
    • SMAX Missing monitoring metrics for CMS/Native SACM
      • CMS not fully ready for auto-healing (rolling restart takes more than 1 minute)
      • SMAX Missing correlations on several S1 / S2 alerts, for example, 5xx errors, and soft interrupts.
      • 5xx errors: put errors into categories - Soft interrupt: more metrics / diagnostics to get the detailed breakdown of the interrupt
      • SMAX It is missing the overall throttling mechanism/rate limit which causes unexpected outages on the farm.
      • 24.4?
      • OO OO upgrade takes hours to finish, the OORAS pods can only be upgraded one after another.
      • Solved in 24.3.
  • Security
    • OPS Missing the WAF rules rolling out, the farm is visited by malicious requests every day
      • WIP
      • OPS Major security KPI missing, including Qualys score, SIEM integration, etc.
  • Compliance
    • Missing the standard/certification for EU-managed
  • Maintenance & Operation
    • OPS Operation efforts increase when there are more farms, including upgrades, patches, etc (Automation rate is low.)
      • OPS Monitoring need to be improved. More meaningful alerts, less false alert.
      • Aligning the CPU alert threshold to 99.9%.
      • OPS Troubleshooting takes lots of time.
      • OPS Cannot always leverage Ops from other regions / how to grow up Ops from other regions
      • OPS Too many threads, need all the members to do self-driven.
      • OPS Need an option to better to utilize Shen Weis time
      • SMAX Logging issue
      • Accumulated logs cost more - Too much logs slows down troubleshooting - Too much log writing used up the network throughput
      • OO The tenant import feature cannot handle integrations like nativeSACM and OO.
      • SMAX OO Too many special settings to keep the system stable, and many of them can be lost during upgrade.
  • Cost
    • SMAX HCMX FINOPS OO When customer usage increases the resource doesn't increase linearly.
      • SMAX HCMX FINOPS OO FinOps, SMAX, and OO consume lots of resources, CMS resource usage is OK
      • The sizing of HCMX, OO are based on tenant number instead of usage. Usually for almost all the ESM farm, OO need to be medium profile ($65K/y) or even larger, which doesn't contribute any license revenue. - FinOps cost is usually more than SMAX large profile ($113K/y) - SMAX sizing is not helpful for medium or large sized customers. Usually the farm need to double or triple the resource required by sizing guide. - There is no sizing guide for integration, including API integration, nativeSACM, etc.