Trial environments for enterprise prospects, kept away from production
Plan an enterprise software evaluation with agreed workflows, separate access and integrations, controlled changes, and a recorded closeout.
Customers send database work, backups and restores to our engineers as tickets. Each guide shows how that work gets done, including when you’re better off staying on a major public cloud.
32 guides
Plan an enterprise software evaluation with agreed workflows, separate access and integrations, controlled changes, and a recorded closeout.
Development & QA
What two QA stacks have to keep apart when a hotfix and the next release test in the same week: the database, sandbox accounts, sign in callbacks and scheduled jobs.
Development & QA
The five items in a regression record for one release: build, environment, dataset, results and approval, with the public framework that asks for them.
Development & QA
Set a soak test's length from the slowest scheduled cycle in the system, hold it at average load, and read memory, connections, disk and latency as slopes.
Operations
Count overlapping demos on the calendar, check whether one tenant can be reset alone, and multiply stacks by releases before choosing a demo environment pattern.
Development & QA
A ledger for one regression cycle: hand rebuilds, dataset loads, fixture setup, triage and reruns, with the commands that show which hours the tests own.
Development & QA
Set a release's latency budget from the run to run spread, measure the last release and the candidate on one environment, and count queries per request.
Development & QA
Move QA off AWS first when it rebuilds often from data staged once and none of its tests call AWS services. Price the path back before you move.
Planning
What cloud repatriation takes for a team under 50 engineers: who replaces a failed disk, how to sort the bill, and why QA moves before production.
Planning
Replay one outage across a PaaS, AWS or Azure with a consultant, and a managed private cloud to see whose engineer does each operating step.
Operations
The vendor's supported platform list decides where a legacy application runs. Three options as SQL Server 2016 and Windows Server 2016 age out.
Planning
AWS Business Support+ costs $3,000 a month on a $40,000 single-account bill, and more across accounts or with a prepaid reservation.
Operations
Questions for a managed private cloud provider, grouped by the failure each one guards against, with the evidence a strong answer includes.
Migration
Recount a Broadcom renewal quote for three ESXi hosts from the cores you run, and price the replacement hosts before choosing a VMware alternative.
Planning
Ten TB of AWS internet egress lists at $912.60. With NAT Gateway, cross-zone traffic and replication, one 10 TB month came to $1,349.07.
Migration
Windows Server 2016 end of support lands January 12, 2027. Give each server role its own action and cutover weekend, and know where ESU applies.
Migration
SQL Server 2016 end of support passed on July 14, 2026. Rerun the heaviest cycle on a restored copy running 2022 or 2025 before the upgrade.
Planning
Sort an Azure bill increase into added work, leftover resources and rate changes in about an hour in Cost Management, then act on each one.
Development & QA
What a QA configuration pins above the golden image, who approves each version, and the order Petrichor builds a stack from it, with a check after each layer.
Operations
Which servers need a backup twin, how the database decides who's in charge, and how to time a failover drill.
Development & QA
Five operating systems and five databases make 25 cells, and vendor platform tables cut them to 21. Decide which run on every build, and where.
Development & QA
Collect the customer's patch levels in the first reply, keep busy releases built, and give each stack a known dataset and a test license.
Migration
Time each cutover step against a restored copy of production, and fix the clock time for the rollback decision before cutover night.
Planning
Price one application's AWS or Azure cost with its support fee, open commitments and engineer hours, then find the quote a provider must beat.
Operations
Turn each database's recovery targets into backup settings, then time a restore that shows whether a host can meet them.
Development & QA
Dev, QA and staging each answer a different question. Here's how to set up each one so its tests mean something.
Operations
Before you pick a provider, write down who does each task, who signs off, and how you'll know it got done.
Planning
Four checks for an application on AWS or Azure: the managed services it calls, the versions it pins, shared hardware behavior, and ticket owners.
Development & QA
Choose a reset method by the state it leaves untouched, and check the environment against a written starting state before the next run.
Operations
What a software vendor and its hosting provider write down before a release: who can stop it, the artifact digest, and the schema change.
Planning
Five questions to answer before real customers use an app you built with AI tools: where it runs, who can get in, backups, patching and cost.
Planning
A plain walk through a cloud invoice for business owners: the five lines that carry most of the cost, and the cost the invoice leaves out.