If most of your users are on the East Coast or in Europe, put your server in Ashburn. If they’re spread across the US and Canada, or you’re building around Chicago’s futures markets, put it in Chicago. And if an outage would really hurt, run both: primary where your users are, standby in the other city.
That’s the answer. The rest of this post is the data behind it, so you can check it against your own traffic.
Round-trip times from each metro
These are our own numbers. On September 30, 2026, our network team had about 15 RIPE Atlas probes in each city, on different ISPs, ping our Chicago, Ashburn and Los Angeles sites. Each figure is the median of those probes’ average round trip, so it’s close to what a typical user in that city sees when they reach us. The raw results are public for Chicago, Ashburn and Los Angeles.
| Users in | To Chicago | To Ashburn | To Los Angeles |
|---|---|---|---|
| New York | 19 ms | 7 ms | 64 ms |
| Boston | 31 ms | 21 ms | 73 ms |
| Atlanta | 24 ms | 17 ms | 46 ms |
| Miami | 41 ms | 29 ms | 60 ms |
| Toronto | 14 ms | 25 ms | 61 ms |
| Montreal | 24 ms | 15 ms | 67 ms |
| Dallas | 24 ms | 33 ms | 33 ms |
| Denver | 30 ms | 45 ms | 34 ms |
| Los Angeles | 48 ms | 62 ms | under 1 ms |
| Seattle | 44 ms | 65 ms | 28 ms |
| Vancouver | 48 ms | 67 ms | 31 ms |
| Winnipeg | 20 ms | 41 ms | 60 ms |
| Calgary | 38 ms | 56 ms | 52 ms |
| London | 93 ms | 80 ms | 134 ms |
| Frankfurt | 107 ms | 89 ms | 147 ms |
Winnipeg and Calgary had fewer probes (6 and 8), so read those two rows as rough. Between our own sites, Chicago to Ashburn is about 24 ms and Chicago to Los Angeles about 40 ms over the point-to-point circuits that link them.
Microsoft’s Azure round-trip latency table tells the same story with a full month of data (median values for the 30 days ending July 30, 2026). East US, in Virginia, sits 19 to 20 ms from North Central US in Illinois. Traffic from UK South takes 78 ms to reach Virginia and 97 ms to reach Illinois, while traffic from West US 2 takes 73 ms to Virginia and 51 ms to Illinois.
Two things jump out. Ashburn wins every East Coast and European metro, by 7 to 18 ms. Chicago’s worst US number in the table is Los Angeles at 48 ms, while Ashburn needs 62 ms or more to reach Los Angeles, Seattle or Vancouver.
So within North America, Chicago has the better worst case (48 ms against 67). Ashburn has the better best case (7 ms against 14).
These are medians across a lot of different ISPs, and one user’s number can land a few milliseconds either side, depending on how their provider routes. That’s the reason to test from your own users’ networks instead of trusting any single table, this one included.
Why Ashburn is the East Coast default
Ashburn is where networks meet. PeeringDB lists 347 networks on the Equinix Ashburn exchange, against 253 on Equinix Chicago. More networks on the same fabric means more of your traffic reaches users’ ISPs directly instead of taking an extra transit hop.
Europe is close, too. Transatlantic cables such as MAREA and Dunant land at Virginia Beach and carry traffic north to the Ashburn cluster, which is why London users reach our Ashburn site in about 80 ms, against 93 ms to Chicago.
Then there’s the cloud. AWS us-east-1 is in Northern Virginia, Google’s us-east4 region is in Ashburn, Oracle runs us-ashburn-1, and Azure’s East US is in Virginia. If you run hybrid, say a database on bare metal with some services in AWS, being a few milliseconds from us-east-1 matters more than anything else on this list.
GigeNET’s Ashburn servers sit in Equinix DC3 in Loudoun County, per our Ashburn dedicated servers page. Our data center locations page lists that facility as SOC 2 Type II and ISO 27001 certified, which helps when you’re filling out security questionnaires for healthcare or finance clients.
Why Chicago wins on national reach
Chicago is the middle of the map. Denver is 30 ms from Chicago and 45 ms from Ashburn; Seattle is 44 ms against 65 ms. If your users are scattered across the country and you only want one server, Chicago gives the fewest people a bad experience.
Canada leans Chicago as well, with Montreal the exception. Toronto is 14 ms from Chicago and Winnipeg about 20 ms, while Calgary is about 38 ms from Chicago against 56 ms from Ashburn.
Finance is the other draw. CME Group’s Globex matching engine runs in the CyrusOne data center in Aurora, Illinois, about 30 miles west of downtown, as Databento’s colocation write-up explains. Be honest with yourself here. If your strategy needs microseconds to that engine, the only answer is rack space in Aurora itself, and no dedicated server anywhere else in the metro will change that.
The cloud neighbors are thinner but real. Azure’s North Central US region is in Illinois, Oracle runs us-chicago-1, and AWS has Local Zones in Chicago that hang off us-east-1. Google’s closest region, us-central1, is in Council Bluffs, Iowa.
GigeNET’s Chicago facility is in Arlington Heights, IL. The locations page calls it our largest facility and home base, with a DWDM fiber ring and a 24/7 staffed NOC on site. The Chicago dedicated servers page lists Cogent, GTT, PCCW and Zayo as carriers. We also run a Los Angeles location for West Coast and Asia-Pacific traffic. It’s 28 ms from Seattle and 31 ms from Vancouver in the table above, which changes the math if your users are mostly on the Pacific side.
Power and disaster risk
The risk worth watching in Northern Virginia is the grid. On July 10, 2024, a fault on a 230 kV line caused about 60 data centers, drawing roughly 1,500 MW, to switch to backup power at the same moment. NERC’s incident review treats it as a warning about how much load is packed into one region.
Those facilities rode through on UPS and generators, which is exactly what they’re built for. For someone renting a server, the bigger practical risk is software concentration: the AWS us-east-1 disruption that began late on October 19, 2025 ran about 14.5 hours start to finish by Amazon’s own summary, and plenty of apps hosted nowhere near AWS still broke because something they depended on lived there.
Chicago has its own weather, from tornado season to hard winters. Neither city is risk-free. The win is diversity: two metros about 24 ms apart, on different utilities, rarely have a bad day together.
The decision rule
- Mostly East Coast or European users: Ashburn. You save roughly 7 to 18 ms on every round trip to your biggest audience, and you’re next door to us-east-1.
- National or Canadian audience, or Chicago-exchange work: Chicago. You give up about 12 ms on New York and win 15 to 21 ms on Denver, Seattle and Vancouver, plus about 9 ms on Dallas.
- Can’t afford downtime: active-passive across both. Run the primary where most users are and a warm standby in the other city, with asynchronous database replication. At about 24 ms apart the replication lag stays small, but synchronous replication would add roughly 24 ms to every commit, so don’t.
I’d skip active-active across two metros unless you already have the team to run it. Split-brain bugs cost more than the milliseconds you’d save.
Price shouldn’t decide this. Our public order forms start at $126 a month in Chicago and $159 a month in Ashburn, but the two entry builds aren’t the same machine, so compare the exact configuration you need. If you already own hardware, colocation can undercut both; our 36-month colo vs dedicated breakdown runs the numbers for Chicago.
Before you commit, test. Ten minutes with mtr from your office or a customer’s ISP to each of our sites beats any table.
Not ready to order yet?
Tell us what you’re after and we’ll email you the build and the price. Need it today? Call (800) 561-2656 and we’ll check what’s in stock.
FAQ
Close enough for planning. Ashburn is roughly 30 miles northwest of the city, and that distance adds well under a millisecond of round-trip time. The bigger differences come from which carriers and exchanges a facility is connected to.
More than it looks. A new HTTPS connection needs three or four round trips before the first byte of the page arrives, so a 20 ms gap can turn into 60 to 80 ms on first load. For games, VoIP and anything real-time, people feel the difference directly.
You can run trading applications on one, but you won’t win speed races. Databento puts the fiber path from Aurora to the downtown Equinix buildings at no less than 287 microseconds one way, and firms colocated in Aurora skip that entirely. A Chicago-area server is a good fit for market-data processing, risk checks and order management that care about milliseconds. Spot FX is a different map, with most retail brokers’ trade servers in New Jersey or near London; our forex VPS latency guide covers where each one sits.
Then neither city is ideal on its own. Chicago reaches Seattle in 44 ms and Los Angeles in 48 ms, which is fine for most web apps but noticeable for games and voice. If the Pacific side is a big share of your traffic, look at our Los Angeles location as the primary or as a third site.
Yes. Our dedicated server pages say all three GigeNET locations can be interconnected over a private backbone, so replication traffic doesn’t have to cross the public internet. Mention it when you ask for a quote.
If you already know where your users are, start with the city page above that matches. If you want a pair across both metros, send the workload through our custom quote form and we’ll price the active-passive setup.