SLA and Reliability Planning: Uptime, Downtime, and TCP Sizing

Reliability commitments and TCP configuration both hinge on translating an abstract target (like "99.9% uptime") into a concrete, checkable number.

MSS: the actual payload size in a TCP segment

Maximum Segment Size is MTU minus header overhead. Standard Ethernet's 1,500-byte MTU minus 20 bytes each for IP and TCP headers gives an MSS of 1,460 bytes — the number you'll see constantly in packet captures, since it's the largest chunk of application data that fits in one unfragmented TCP segment.

Uptime percentage: turning downtime into a percentage

A 30-day month (43,200 minutes) with just 15 minutes of downtime works out to 99.9653% uptime — comfortably above the common "three nines" (99.9%) SLA tier.

Allowed downtime: the reverse calculation for planning

Working backward from a 99.9% target over that same 30-day month allows about 43.2 minutes of downtime — each additional "nine" of reliability cuts that allowance by roughly 10×, which is why 99.99% only allows about 4.3 minutes per month.

DNS TTL: planning propagation time before a change

An 86,400-second (24-hour) TTL means cached DNS records can take up to a full day to update everywhere after a change — lowering TTL in advance of a planned migration is standard practice specifically to shrink this propagation window.

VPN overhead: the throughput cost of encryption

A 100 Mbps connection with 12% typical VPN overhead delivers roughly 88 Mbps of effective throughput — the gap between raw connection speed and what you'll actually see through the tunnel.

Turning reliability targets into real numbers

SLA tiers, TCP segment sizes, and DNS propagation windows all sound abstract until you calculate the actual minutes or bytes involved. Try the MSS calculator, uptime percentage calculator, allowed downtime calculator, DNS TTL calculator, and VPN overhead calculator.