TLS/SSL Handshake Overhead Calculator
Estimates the additional latency and data overhead introduced by TLS/SSL handshakes based on protocol version, round-trip time, and session parameters.
Network latency between client and server (one round trip).
TLS 1.3 requires fewer round trips than older versions.
Session resumption reduces handshake round trips.
Typical single cert ≈ 2–4 KB; chain with intermediates ≈ 4–8 KB.
Connection bandwidth used to estimate transmission delay for cert data.
Number of new TLS connections established per second.
Results will appear here.
Formulas Used
Handshake Latency (ms)
Handshake Latency = NRTT × RTT
where NRTT = number of round trips required by the chosen protocol and resumption mode.
Certificate Transmission Delay (ms)
Cert Delay = (CertSizeKB × 8) / BandwidthMbps
Total Per-Connection Overhead (ms)
Total Latency = Handshake Latency + Cert Delay
Handshake Data Volume (bytes)
TLS 1.3 Full: ≈ 790 + CertBytes
TLS 1.2 Full: ≈ 826 + CertBytes
Resumed (any): ≈ 492 bytes (no certificate transmitted)
Aggregate Bandwidth (Mbps)
BW = (ConnPerSec × HandshakeBytes × 8) / 1,000,000
Round-Trip Counts by Mode
| Version | Full Handshake | Resumption | 0-RTT |
|---|---|---|---|
| TLS 1.0 / 1.1 | 2 RTT | 1 RTT | N/A |
| TLS 1.2 | 2 RTT | 1 RTT | N/A |
| TLS 1.3 | 1 RTT | 1 RTT (PSK) | 0 RTT |
Assumptions & References
- RTT counts follow RFC 8446 (TLS 1.3) and RFC 5246 (TLS 1.2) handshake message flows.
- TLS 1.3 reduces the full handshake from 2-RTT (TLS 1.2) to 1-RTT by merging key exchange into the first flight (RFC 8446 §2).
- 0-RTT (Early Data) in TLS 1.3 allows application data in the first flight, eliminating handshake latency for resuming sessions, but is susceptible to replay attacks (RFC 8446 §8).
- Certificate sizes: a single end-entity certificate is typically 1–3 KB; a full chain with intermediates is 4–8 KB (Let's Encrypt, DigiCert documentation).
- Handshake byte estimates are approximations based on typical ECDHE-RSA cipher suites; actual sizes vary by cipher suite, extensions (SNI, ALPN, OCSP stapling), and certificate size.
- Crypto CPU overhead estimates (~0.3–0.5 ms for ECDHE P-256) are illustrative; actual values depend on hardware, HSM use, and server load.
- Session resumption via session tickets (RFC 5077) or PSK (RFC 8446 §2.2) avoids full certificate exchange, reducing both latency and bandwidth.
- References: RFC 8446 (TLS 1.3), RFC 5246 (TLS 1.2), RFC 5077 (Session Tickets), Cloudflare TLS Performance blog, Google QUIC/TLS research.