01 / NetworkingFoundations
A name is a starting point.
When you open a website, its name helps your computer find an address. DNS is the directory that answers this question. It does not, by itself, decide which page appears or what an application does after the connection opens.
Name→DNS answer→Connection→HTTP request
One address, more than one website
Several names can resolve to the same address. A web server can use the hostname in the request to choose the right site. For HTTPS, the TLS handshake commonly also carries a server name, called SNI, so the server can select the appropriate TLS configuration.
These steps are related but distinct. A successful DNS answer does not prove that a service is running, and an open port does not prove that a website returns the expected content.
A useful troubleshooting order
First confirm the name resolves to the intended address. Then check whether the connection can open, whether the certificate matches the requested hostname, and whether the HTTP response contains the page you expected. Changing one layer at a time makes the result easier to explain.
02 / ReliabilityPractice
A backup is a future restore.
A copied directory is only the beginning. A useful backup must contain the data needed to reconstruct an application, and that data must represent a consistent state. A database being written while its files are copied can require special handling.
Ask three concrete questions
- What is included? Keep a short inventory of data, configuration, and any credentials required for recovery.
- Where else does it live? A copy on the same disk may help after an accidental edit, but it cannot protect against losing that disk.
- Can it be restored? Try a restore in an isolated location and verify the application can read the recovered data.
A checksum can confirm that a file has not changed relative to a recorded value. It cannot prove that the backup contains everything the application needs.
Make a small recovery note
Record when the copy was made, the application version, the restore steps, and the result of the last restore test. Protect backups containing credentials as carefully as the original data. A concise, tested procedure is more valuable than a long document nobody has tried.
03 / The webTrade-offs
Fast answers need an expiry date.
A cache keeps a previous result so it can be reused. Browsers cache files, DNS resolvers cache answers, and applications cache expensive computations. This saves time and work, but the saved answer may become older than the underlying data.
Freshness is a choice
Longer lifetimes reduce repeated requests. Shorter lifetimes let changes become visible sooner. The right choice depends on the content: a versioned image can usually remain cached longer than information that changes frequently.
For web assets, a filename tied to the file's content is a useful pattern. A new version gets a new URL, while the old version can retain its long cache lifetime. HTML often needs a different freshness policy because it points to those assets.
Why a refresh may not show a change
There may be several caches between a browser and the source. Inspect the response headers and identify which layer supplied the result before clearing caches at random. Also check whether the HTML still references the previous asset URL.