Buy the serious machine, the way we have since 2004.
Twenty-two years of buying the serious machine instead of the adequate one is the reliability argument. Receipts, not nostalgia.
In this article 4 sections
- The doctrine has two lines: own your equipment or resell someone's promises, and cheap, reliable, fast: pick two. We always picked the same two.
- Twenty-two years, one decision, repeated: quad-socket servers in 2004, Fibre Channel done properly in 2012, our own network in the 2020s, HPE 3PAR all-flash federation in Athens in 2026.
- What it buys you today: our platform is built so that hardware failure does not reach your data. Hardware fails; the platform is built so you stay online while we replace the part.
Standing at the counter with two parts in front of you, you either buy the serious machine or the adequate one. We have been making that purchasing decision since 2004, and we have made it the same way every single time.
The doctrine, plainly
Two sentences have governed every purchase since. The first: in this industry, you own your equipment or you resell someone's promises. Servers, storage, network hardware, all of it. Owning the equipment is what lets us stand behind the reliability we sell. The second is older than the internet, and we never argued with it.
“Cheap, reliable, fast. You can have two of the three.”
We picked reliable and fast, and we put the specs on the page so you could check.
Twenty-two years of the same decision
The doctrine is easy to state and hard to keep, because keeping it means choosing the serious part every time silicon turns over. Here is what that looked like across four eras.
The quad-socket start
Built in Amsterdam on HP ProLiant quad-socket servers with battery-backed RAID cache. Serious machines from the first day.
Fibre Channel, properly
Virtual machines on EMC Fibre Channel storage, running banks and airlines. We kept it in internal beta for 18 months before a customer touched it. HP blades and Brocade fabrics arrive the same era and never leave.
Our own network
The center of gravity moves to Athens and the network is rebuilt edge to spine: our own autonomous system, bilateral peering at GR-IX, Cisco ASR 9000 series edge, Nexus 9000 leaf-spine, a real per-tenant fabric on our own silicon.
The Athens cage
A new cage at Digital Realty Athens: HPE 3PAR all-flash storage running active-active Peer Persistence in-site over dual Brocade fabrics, with asynchronous cross-site replication to Amsterdam, and HPE and Dell compute. The machines in the current photos on this site live here.
What it buys you today
Add up twenty-two years of choosing the serious part and it buys one specific thing. Hardware fails; the platform is built so that a failure does not reach your data, and you stay online while we replace the part. The spare lives onsite.
HPE 3PAR Peer Persistence in-site, plus async replication to a second site
Spares on the shelf, onsite
A named engineer on the call
The part you can't buy off a spec sheet
There is one line on our about page, first written in 2012 and still there, that no purchasing decision produces on its own: we love this technology the way a car person will spend an entire weekend polishing a valve cover, because it has to be perfect. You cannot order that from a distributor. It is why the machines are labeled and photographed, why every cable in the rack has a name, and why the same decision got made correctly for twenty-two years.
Straight answers
Aren't other hosts cheaper?+
We build for reliability first.
What happens when hardware fails?+
Our own engineers swap the part from onsite spares rather than filing a ticket with a vendor depot, and the platform is built so you stay online through it, because the storage is federated and the fabric is doubled.
Can I visit the machines?+
Yes. Three datacenters with real addresses: Athens, Amsterdam and Toronto. Ask and we will walk you through; the machines in the current photos on this site are the actual machines.
Buy the serious machine. It is the whole argument, and it is the argument we have been making with hardware, not adjectives, since 2004. Talk to an engineer when you want to size yours, or come read the full pedigree on the about page and see where the receipts lead.
Come stand in front of the machines.
Real addresses, escorted visits for any customer who asks. The pedigree, the cage, and the people who rack it.
See the pedigreeYou’ve reached the end.
Put the reading to work
What are
you running?
Tell us about the workload, traffic and recovery requirements. Start with the server builder, or talk through the details with an engineer.
Written by the engineers who build and run the infrastructure it describes.Talk to an engineer