A breakdown of why I chose 123Net DC1 in Southfield, Michigan: the facility, the network, the Detroit Internet Exchange, and what it actually means for VPS latency.
Esta publicación solo está disponible en inglés por el momento.
This post covers why we picked that facility, what the network looks like, and what the real-world numbers mean for customers.
The Facility: 123Net DC1, Southfield MI
123Net DC1 is an 80,000 square foot data center with 20 megawatts of available utility power. Tier IV means fully fault-tolerant with no single point of failure for power or cooling. The facility holds SOC 2 Type II, HIPAA, and PCI-DSS certifications. The building is reinforced concrete, not in a flood plain, and has five entrance vaults for physical security and redundant fiber entry.
Those specs matter for one reason: uptime. A Tier IV facility with 20+ carriers on-site and redundant everything is the kind of place where planned maintenance does not require downtime. We did not want to colocate in a building where a single UPS failure or one fiber cut takes everything offline.
There is also a practical reason for the Detroit metro specifically. 123Net owns and operates its own statewide fiber network. That means we are not dependent on a third-party carrier for connectivity within the building. The upstreams we use transit through their own infrastructure before hitting external peers.
The Detroit Internet Exchange (DET-iX)
The most important thing about 123Net DC1 is what lives inside it: the Detroit Internet Exchange, also known as DET-iX. It is one of the largest fee-free internet exchanges in the world, operating at 2.5 terabits per second of peak traffic across more than 370 BGP sessions.
The member list includes Apple, Microsoft, Google, and 90+ other peers. RackWorks is not a Detroit-IX member and does not peer on the exchange yet; our connectivity comes from a Detroit-IX connected upstream. When traffic from my VMs needs to reach a Google server or an Akamai CDN node, our upstream can hand it to that network across the exchange, inside the building, instead of routing it out through a transit carrier.
That is what a colocation customer actually buys when they pick a well-peered facility. It is not just rack space. It is proximity to the exchange.
Latency: Real Numbers from the Looking Glass
We run a looking glass at rackworks.net/lookingglass.php so customers can check latency before buying. The numbers below are from actual ICMP probes, not marketing estimates.
| Destination | Round-Trip Time |
|---|---|
| Chicago, IL | 10 ms |
| New York, NY | 16 ms |
| Toronto, Canada | 22 ms |
| Dallas, TX | 28 ms |
| Seattle, WA | 32 ms |
| Miami, FL | 43 ms |
| Los Angeles, CA | 47 ms |
| Amsterdam, NL | 97 ms |
| Tokyo, Japan | 146 ms |
Chicago at 10ms is as good as it gets for a Detroit-area host. The Midwest and Great Lakes region, including Toronto, Cleveland, Cincinnati, and Indianapolis, is all sub-30ms. East Coast is under 20ms. The intercontinental numbers are what physics allows; there is no optimizing past the speed of light.
If your use case is algorithmic trading against Chicago or New York exchanges, latency-sensitive gaming for a Midwest player base, or a VPN exit node for US traffic, those numbers are competitive with anything hosted in the region.
The Hardware
Our servers run Dell PowerEdge hardware. Dual Intel Xeon Gold 6138 CPUs, 20 cores each, hyperthreaded, giving 80 threads total. Storage pools are split between NVMe for the KVM and LXC SSD plans and SATA SSD for the budget tiers.
These are purpose-built 1U rack servers with no consumer compromises. ECC RAM, redundant PSUs, iDRAC for out-of-band management. If a DIMM fails at 3am, the monitoring fires an alert and it can be replaced without touching the VMs.
What This Means for Customers
The short version: you get enterprise-grade colocation infrastructure and serious network connectivity without the enterprise price tag.
Most budget VPS providers use mid-tier facilities with limited peering. They oversell CPU and RAM to hit a price point. Traffic goes through one or two upstream carriers with no exchange presence, which means extra hops and worse latency to major networks.
Hosting on RackWorks means your VM is physically inside the same building as DET-iX, one of the largest internet exchanges in North America. Our upstream is Detroit-IX connected, so traffic to networks on the exchange, including Google, Microsoft, Apple, and 90+ other peers, can be handed across it locally instead of through a transit carrier. Your latency to Chicago is 10ms, not 30ms. That is the network behind our Detroit VPS hosting plans.
When something goes wrong, it gets fixed. No ticket queue, no Tier 1 support reading from a script. Direct access to the team that built and operates the infrastructure.
If that sounds like what you need, the KVM VPS plans and LXC container plans are on the main site.