Calculator Inputs
Example Data Table
| Scenario | Objects | RTT | Bandwidth | Mode | Expected Benefit |
|---|---|---|---|---|---|
| Small site | 12 | 40 ms | 50 Mbps | Serial reuse | Moderate |
| Media page | 48 | 90 ms | 20 Mbps | Multiplexed reuse | High |
| Cached repeat visit | 35 | 70 ms | 30 Mbps | Pipelined reuse | Medium |
Formula Used
Total requests = 1 + embedded objects × (1 − cache hit rate)
Payload bytes = HTML bytes + effective embedded objects × average object bytes
Transferred bytes = compressed payload bytes + total header bytes
Transfer time = transferred bits ÷ bandwidth
Non persistent time = DNS + connection groups × setup cost + request cost + transfer time
Persistent time = DNS + reused connection setup + persistent request model + transfer time
Savings = non persistent time − persistent time
How to Use This Calculator
- Enter the number of embedded objects on the page.
- Add the HTML size and average object size.
- Enter bandwidth, RTT, DNS, TCP, and TLS values.
- Choose the persistent request model.
- Adjust cache and compression assumptions.
- Press Calculate to compare both delivery models.
- Use CSV or PDF buttons to save the result.
Understanding Persistent HTTP
Understanding Persistent HTTP
Persistent HTTP keeps a TCP connection open after one response. The browser can send more requests through the same path. This reduces repeated handshakes. It also lowers latency on pages with many files. Images, style sheets, scripts, fonts, and data calls all benefit. A single page may contain dozens of objects. Without reuse, each object may need another connection. That creates extra waiting before useful bytes move.
Why This Calculator Matters
This calculator estimates the time difference between repeated connections and reused connections. It combines object count, round trip time, transfer size, bandwidth, server processing, TLS cost, cache hits, compression, and parallel connection limits. The result is not a packet capture. It is a planning estimate. Still, it helps teams see which factor dominates. High RTT networks gain often. Small files gain too, because handshake time can exceed transfer time.
Reading the Results
The estimated non persistent time shows a page model with separate connections. The persistent time shows a model with connection reuse. The savings value shows milliseconds avoided. The percentage shows the relative improvement. The transfer time is separated from connection overhead. This makes the result easier to audit. A negative saving can appear if inputs model an overloaded persistent path. Check request mode, stream limit, and requests per connection when that happens.
Practical Use
Use realistic inputs from logs, browser tools, or synthetic tests. Measure RTT from the user region, not only from the server room. Include TLS handshakes for secure sites. Adjust cache hits when repeat visitors load the page. Use compression when text assets are compressed. Keep object count close to real pages. Then compare several scenarios. You can test high latency mobile users, office users, or overseas visitors. The calculator also helps explain why modern delivery uses keep alive, multiplexing, caching, and compression together.
Optimization Notes
Persistent connections are most useful when many requests share one origin. They do not remove download time. They remove repeated setup time. Good servers also tune keep alive limits. Very long limits may waste memory. Very short limits may force reconnection. Balance matters. Test with real traffic patterns before changing production settings. Review error rates after changes, because speed should never reduce reliability under load.
FAQs
What is persistent HTTP?
Persistent HTTP reuses an open connection for multiple requests. It reduces repeated TCP and TLS setup work, which can improve page loading time.
What does this calculator compare?
It compares estimated non persistent delivery time against persistent delivery time. It also shows savings, transferred data, connection count, and improvement percentage.
Should I include the main HTML request?
No. Enter only embedded objects. The calculator automatically adds one request for the main HTML document.
What is RTT?
RTT means round trip time. It is the time needed for a network signal to go from client to server and back.
Why does TLS matter?
TLS can add handshake delay before secure data is exchanged. Reusing a connection can avoid repeating that cost for every object.
What does multiplexed mode mean?
Multiplexed mode estimates multiple request streams sharing one connection. It is useful for modern connection reuse models with concurrent delivery.
Can this replace real performance testing?
No. It is an estimation tool. Use it for planning, teaching, and comparison. Confirm production decisions with real measurements.
Why can savings become negative?
Negative savings may appear when the selected persistent model is constrained. Review stream limit, request mode, object count, and connection reuse settings.