Editing X Window System (section)

===Client–server separation===
X's design requires the clients and server to operate separately, and device independence and the separation of client and server incur overhead. Most of the overhead comes from network [[round-trip delay time]] between client and server ([[Network latency|latency]]) rather than from the protocol itself: the best solutions to performance issues depend on efficient application design.<ref>{{Cite web |last=Packard |first=Keith |title=An LBX Postmortem |url=https://keithp.com/~keithp/talks/lbxpost/paper.html |url-status=live |archive-url=https://web.archive.org/web/20231115105104/https://keithp.com/~keithp/talks/lbxpost/paper.html |archive-date=Nov 15, 2023 |website=keithp.com}}</ref> A common criticism of X is that its network features result in excessive complexity and decreased performance if only used locally.

Modern X implementations use [[Unix domain socket]]s for efficient connections on the same host. Additionally [[shared memory]] (via the [[MIT-SHM]] extension) can be employed for faster client–server communication.<ref>{{Cite web |last=Corbet |first=Jonathan |date=1991 |title=MIT-SHM—The MIT Shared Memory Extension How the shared memory extension works |url=http://www.xfree86.org/current/mit-shm.html |url-status=live |archive-url=https://web.archive.org/web/20231222105026/http://www.xfree86.org/current/mit-shm.html |archive-date=Dec 22, 2023 |website=XFree86}}</ref> However, the programmer must still explicitly activate and use the shared memory extension. It is also necessary to provide fallback paths in order to stay compatible with older implementations, and in order to communicate with non-local X servers.