Agent Guide to support working with: STARLIMS Performance Triage

Performance tickets are the hardest tickets we get: the complaint is vague ("STARLIMS is slow"), the cause can sit in the client, the network, IIS, the database, the customer's own code or in load we never see, and the people reporting it are usually not the people who can answer our questions. The customer-facing article STARLIMS Performance Triage fixes the customer side of that: it gives them, in one place, the questions, the requirements and the data collection we would otherwise ask for piece by piece over weeks. This guide is the agent side. It tells you which chapter to send when, what to do with what comes back, and when to stop and bring in our DBA. Working it in order protects both the customer and us from the two classic failure modes: tuning a system that doesn't meet the basics, and collecting mountains of data nobody interprets.

How to use the customer facing article

Never send the whole article. Point the customer to it once and, in each reply, name the chapter(s) to complete now. Keep one tracked list of findings and recommendations in the ticket from the first reply on; a long performance ticket loses suggestions otherwise.

VERY MUCH WORK IN PROGRESS!!!

1. Intake;
   1. Understanding the problem
      Chapter 1 from the Confluence document and instead of copying the questions into a ticket, this becomes chapter 1 in a customer facing article about Performance Triage and the customer is requested to answer those questions in the performance ticket.
   2. Checking the basics (for on-prem) customers
      In the 'STARLIMS Installation & Performance Checkup' document, I do a lot of configuration checks. Is STARLIMS / SQL Server / Oracle / etc. setup as we suggest it. Performance tuning won't work if the basic setup is not correct.
      Chapter 2 in the customer facing article about Performance Triage will contain all requirements from 'STARLIMS Installation & Performance Checkup' document (extended with the things from more recent TP guides). The customer will be asked to check all of those and report back to us
   3. Getting more data
      In the 'STARLIMS Installation & Performance Checkup' document, I request a lot of information from the customer. When Stefanie does her analysis, she also requests a lot of information. Some things overlap; some things don't. We'll combine this in this step to get what support needs.
      Chapter 3 in the customer facing article about Performance Triage will contain all data requests from the performance doc and the ones from Stefanie that overlap
2. Narrow down the cause;
   1. Process the data of the intake v1
      1. With 1a completed when can describe the problem and communicate back to the customer what the exact problem is we're going to work on
      2. With 1b completed, when can review the basics and already suggest they take appropriate action. Chapter 2 in the customer facing article about Performance Triage will contain (just as in the perf. document) reasons behind certain required configurations.
         Data from 1c might change when suggestions from 1b are implemented. Go step by step!
   2. Process the data of the intake v2
      1. With 1c completed we've got a lot of data from the customer to review.
         A new internal article on Zendesk (Performance Triage (Agent Guide)) will contain explanation on what to do with all the data points received in 1c
   3. Decide to report a cause back to the customer or involve Stefanie for more advanced performance tuning
3. Report cause and action plan, or
4. Report back current findings and involve Stefanie (on-prem) / Henrique (cloud)
   1. Report back the current findings and action plan for support and also the plan to involve our DBA for further investigation
   2. Getting even more data
      We already requested quite some data, but Stefanie needs more. To be discussed: Chapter 4 in the customer facing article about Performance Triage will contain all data requests from Stefanie