How to get the fun stuff back in your data centre The cloud is a fabulous concept. If you want to try something out, or prototype your latest idea, or give yourself a relatively inexpensive disaster recovery setup, get in there and run up a cloud-based installation. There's something that the cloud lacks, though: it's just not fun or cool. Lists of virtual machines in the …
I am currently in the process of ripping out my entire network and getting a new one installed. Part of the UAT I specified was that I'd go around randomly unplugging things and see at what point it fails. It's a genuinely valid methodology for testing (given you think about what you actually want to test for, of course) IMO.
I think a lot of management types forget, overlook or just plain don't understand 2 very good points also raised by this article.
1. IT types generally learn by doing (I sure as hell do)
2. Having IT types who have learned - and know what's what - is extremely good for your business in the long term.
Sadly, this is overlooked in the name of economy (of the false kind) and efficiency (of the false kind). I mean; who wants to waste money and time letting your own people learn by doing when you can get a consultant in and not have to waste a minute or a penny?
...
Right. OK.
A fantastic article, thank you.
[edit for spelling]
Because we've broken them in our labs on purpose and know what the results are...and how to identify the condition, and how to resolve it.
I do have one beef with the article, quoting:
So you have the same firewalls, same family of switches, the same remote management servers, the same fileservers, the same storage devices, and so on at each site...
No, you don't. If you want real redundancy, you have two (or more, it depends on your size) vendors hardware/software in parallel. This will prevent you from having your entire enterprise taken offline by a single vendors' bug.
This will double your support costs, but if uptime and continuity are important, I think you have to do this.
I have 2 of them from 2 different last mile providers (same ISP) coming into the building from different directions. I wanted them coming into different ends of the building but that'd have cost me £50k in construction, so that got nixed.
It's not perfect, but it's better than many people I know.
You have two of them, on separate circuits, trenched, cabled and owned by separate carriers (this IS VERY expensive.) Or you can have two nodes on a SONET ring with two separate building entrances (this IS ALSO VERY expensive.) You do the same with power. Two substations, fed from two separate power grids (again, this IS VERY expensive.)
It is also very nice, as a former bit&byte, home cluster builder (pentium I central control node, full dvd rip in 20 hours with 4 nodes...about as powerfull as your iphone 6S...) Geek, certified for clusters since Compaq was the one doing the nice servers, and turned consultant cum auditor, to go into the DC after the interviews with the playfully IT staff, put a hand on the back of "that" server and say :
"So, according to our interviews, even if I remove this specific power and stop the system it will still be all good"
And look them in the eyes.
N.B. I usually just unplug the Ethernet
N.B.2. Hint : I usually go for the server hosting the quorum disk ^^
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
