The hallmark of a good data centre is, as most people know, redundancy. All major components of the data centre (generators, UPS, cooling) should either be N+1 or 2N, where N = the number of components necessary to keep the servers and critical equipment up and running, with data accessible.
For example if a DC needs 400kw power to run all the servers and critical equipment, N = 400kw. So at the absolute very least, the DC should have backup generators that total 400kw.
So a 2N metric for generators would be 2x 400kw generators. An N+1 metric could take a few forms: if the original 400kw was configured as 2x 200kw gens, N+1 would be 3x 200kw gens, meaning one of the gens could go down and there would still be sufficient power. Or it could be configured as 4x 100kw, then N+1 would be 5x 100kw gens, again meaning one of the gens could go down etc. The same metric applies to UPS and cooling, as well as fiber too.
The other hallmark is concurrent maintainability, which means that there are 2 physical paths for the power to get to the racks, and 1 of those paths can be down for maintenance without affecting the other path. Again the result is better up time / accessibility. All of this takes material investment in power, cooling and fibre equipment, as well as ongoing maintenance of all systems.
Many (but not all) lower priced data centers are not even N+1 (and typically definitely not 2N) across all critical equipment (gens, UPS, cooling) and may not have alternative fiber paths. You need to ask for their equipment map to be sure because many would prefer not to have this conversation. Not investing in the appropriate equipment means they can offer a lower price but the client is assuming much more risk.
Placing servers in a low priced data centre may seem like an excellent way to save money. But it isn’t until you have lost data, or experienced long outages do most businesses understand the folly of that decision.
All of us must have at one time or another, failed to back up at home and then experienced the gut wrench of the dreaded blue screen of a dead hard drive or other hardware malfunction, losing completely or losing access to, valuable data, documents, pictures etc, perhaps been able to get some of it back, perhaps not. Imagine your business experiencing the same thing. Yes it is slightly more costly to be serviced by a quality data center but the regret factor can be huge and the consequences devastating.
The Canadian market has many data centre offerings, with a very wide range of prices. Pricing is largely dependent upon investment in critical infrastructure and reputation. As you move up the price scale you do get the necessary infrastructure but past a certain price point you are merely paying for reputation (and perhaps really nice meeting rooms), but not necessarily increased reliability of access to your data.