When we talk about hardware support contracts with the manufacturer we are being very serious. When we were planning our infrastructure we decided to do something unusual in the server hosting business; we decided to purchase hardware support contracts for all of our hardware to give us 4 hour on site responses by the manufacturer.
It adds some cost so why did we do this? Our customers depend on us. They provide a service to their customers (be they internal or external) and their reputations and ability to continue operating depend on our ability to provide a reliable service. We don’t cut corners. If our infrastructure fails then your customer isn’t going to accept a story like “well, the replacement will be here tomorrow”! We don’t buy support on a couple of bits here and there and avoid the critical components. And we also build in fault tolerance.
Our HP EVA SAN (storage area network) is the heart of the C Infinity private cloud. It’s where all of the data is stored. We use dual path fibre channel HBA’s in the servers, dual switches and dual controllers. The HP blade server chassis has fault tolerant fibre channel and network Interconnects too. The disks have RAID and there’s a hot spare disk. If something does go wrong we “keep on truckin’” and things stay operational at 100% capacity and performance.
I was working late one night last week when I saw an alert on the SAN. After a quick look I saw a disk had failed. Normally RAID would start up with potentially some loss in write performance. However, the disk group had activated the hot spare disk so everything stayed operating at 100%. I immediately logged a call on the HP website. Start the stop watch:
- 20 minutes later the HP support desk rang me to arrange the engineer visit. I arranged escorted access for the visiting engineer to access the required rack.
- In less than an hour DHL had brought a replacement disk to the data centre.
- At exactly 2 hours after I opened the call the visiting engineer called me from the rack. He confirmed with me the disk to replace.
- At 2 hours and 2 minutes, I had added the replacement disk into the disk group and the SAN had started to spread the data back onto it.
So what was the end result for our customers in the private cloud? They didn’t notice anything. There was no drop in performance. There was certainly no data loss. Thanks to our concentration on doing things the right way our customers got the very best from our architecture and HP support.
Please contact us if you’d like to learn how you can get these levels of fault tolerance and reliability in the C Infinity private cloud
Related posts:
