- VP, Product Management
- Posts: 5356
- Liked: 2192 times
- Joined: Jun 05, 2009 12:57 pm
- Full Name: Tom Sightler
I did want to clarify to other forum readers that you statement that you "...get the same exact speed and latency as if the two where in the same room...", while functionally likely true in your case, it is obviously not strictly true from a technical perspective. You certainly have the same throughput, but latency will be increased because of the length of the fiber. With modern fiber and a typical refractive index you will add about 5 microseconds of latency for every kilometer, thus 25 microseconds for your 5K link, and really double that, since each read/write cycle requires a trip each direction, so really 50 microseconds per command. Admittedly not likely to be performance impacting because it's such a small component of the total latency (the disk itself is likely to be 3-4 milliseconds at best so this is maybe 1-2% of the total latency per request), but just trying to be technically accurate.
Keep us updated, sounds like fun! I suspect your breakpoint will be disk long before you hit the limits of any of your individual links.
- Posts: 109
- Liked: 14 times
- Joined: Nov 01, 2011 1:44 pm
- Full Name: Lars Skjønberg
Users browsing this forum: No registered users and 15 guests