Hostile Customer Service / Unreliable Power
I was a new colocation customer with Cloudnium for a couple of months, but ultimately was unable to put my servers into production because I could not get confidence that stable power was being supplied to my equipment.
There are people at Cloudnium who seem to genuinely care about doing a good job. Several of the engineers I dealt with were helpful, and their COO was professional and appeared sincerely interested in resolving the situation. Unfortunately, my experience with one individual, Senior System Administrator Yogesh Pasalkar, ultimately destroyed my confidence in Cloudnium as a production colocation provider.
Before and during my time with Cloudnium, I repeatedly made it clear that I was willing to purchase additional power or move from my 2U allocation to 4U if my equipment required it. I was not trying to operate these servers as cheaply as possible. I wanted a reliable production environment and was willing to pay Cloudnium whatever was reasonably necessary to provide one.
I placed two 1U servers in my 2U allocation. One of those servers operated reliably for weeks while the other experienced unexplained failures. To eliminate my own hardware as the cause, I went so far as to replace the entire server.
The replacement server subsequently experienced the same type of power problem.
At that point I expected Cloudnium to investigate the facility side of the equation: outlet power, PDU, circuit, cabling, power allocation, or anything else that could explain why an entirely different server was exhibiting the same behavior.
Instead, Yogesh took an adversarial position. He suggested that the problem was that I had installed two 1U servers in a 2U allocation and stated that the 2U plan was generally intended for a single 2U server. Cloudnium's COO subsequently clarified that two 1U servers in the allocation were perfectly acceptable and that combined power consumption was what mattered.
More concerning, Yogesh placed the burden on me to demonstrate that the server itself was not at fault without first offering to establish whether Cloudnium was reliably supplying power to it.
At the time, I could not even access the server's out-of-band management because the server did not have power.
This is roughly equivalent to having your car serviced, discovering in the mechanic's parking lot that it will not start, and being told to start the car so that you can prove the engine isn't defective—without the mechanic first checking whether the battery is connected.
The ticket history also documented that I had already replaced the original server and that I had repeatedly offered to purchase additional power or rack space if necessary. That history appeared not to have been considered before responsibility was placed back on my equipment.
This is especially troubling behavior from someone bearing the title Senior System Administrator. In a colocation environment, when both a server and its out-of-band management become unreachable, establishing whether the equipment is actually receiving power should be one of the most basic troubleshooting steps.
My objection is not that Cloudnium experienced a technical problem. Equipment fails. Circuits fail. Data centers have incidents. I would not condemn a provider merely because something went wrong.
My objection is to the troubleshooting and customer-service response when something did go wrong.
When you operate a service business—particularly one entrusted with mission-critical infrastructure—you cannot begin from an adversarial assumption that the customer must prove your infrastructure isn't responsible before you investigate it. Even when the customer ultimately turns out to be wrong, the job is to establish facts and isolate variables.
I made a deliberate effort throughout this process to be patient, kind, flexible, and willing to spend additional money to solve the problem. I even replaced an entire server to eliminate my own hardware as a variable. After the replacement experienced the same problem and Cloudnium still did not inspire confidence that the facility-side power path would be investigated first, I decided to remove my equipment and leave.
I think Cloudnium has the potential to be better than my experience suggests. Their COO handled the escalation professionally, and several members of their technical staff seemed conscientious and genuinely interested in helping. Unfortunately, a colocation provider is only as trustworthy as the response you receive when your production equipment goes dark, and my experience ultimately left me unable to trust Cloudnium with production infrastructure.
