It’s been pretty quiet around here this week, mainly because I’ve been busy. Even more so than usual, which given how busy the summer normally is, is saying a lot.
For the last couple of weeks I’ve been bringing new servers online, moving files, redeploying applications, and generally shifting stuff around as part of a smooth transition to our new network management system. Today, I hit a milestone.
Today, I decommissioned the last remaining RM Community Connect 3 server on our network.
For every person who has to still put up with this antiquated, cobbled-together piece of junk; for every sysadmin who longs to see the back of it; this is for you.
No, I haven’t been sacked – but I can think of someone who should be.
This week, while decommissioning an RM replica DC (which also functioned as a file server, along with about 155 other roles, as RM are fond of doing), I confirmed something I’d suspected for some time.
The RAID controller was not actually configured for RAID.
This entire time, the server has had no fault tolerance on any of its disks. Which is interesting, because when I started at the school, the backups were also not working properly. All of which is made worse by the fact that this server, when it was first installed, was the only server, and was therefore the Active Directory Forest Root.
With no fault tolerance.
Whoever set this up needs to be beaten to death with their own shoes.
One of the advertised benefits to schools of upgrading to Windows 7 has been the improved power management, which when used correctly, will decrease the power usage of computers throughout the school estate. But how is that actually achieved? A lot of people, myself included, assumed that the bulk of the savings would come from better Group Policy support to implement power saving schedules, such as transitioning to Sleep mode after a set time.
While that does help, there are significant benefits even without power management schedules. Simply put, Windows 7 uses less power. I measured the power usage of two computers; identical hardware, but with different software. One ran an RM CC3 build using Windows XP, the other ran Windows 7. Both machines were production workstations with all the normal software I install on them, and they were tested while idling at the logon screen. The machines themselves were 2009 Dell OptiPlex 360 workstations, and the power usage was measured using an in-line mains meter. I did not measure the monitor power usage.
Here’s what I found:
|Windows XP (CC3)||55.0 W||41.1 W|
|Windows 7||45.0 W||1.7 W|
There are two lessons here: firstly, Windows XP is shockingly bad in sleep mode. Given that XP came out in 2001, this shouldn’t be entirely surprising, but I was still astonished by how high the power usage remained even when the machine was supposedly in its low-power mode. To an observer, the two machines were indistinguishable from each other at this time – power LED blinking, fans off, silent operation – but the difference in power usage was outrageous.
The second lesson appears to be that Windows 7 draws less power when you’re not using it. Later on I tested both machines using a CPU stress test, and they pulled the same amount of power, but when left idle, the Windows 7 machine averaged 10W less.
I should note that it’s entirely possible the RM background tasks were responsible for the latter discrepancy; I didn’t have a non-RM XP machine to test. No such difference could explain the sleep mode difference. Similar results emerged on different hardware, so this didn’t appear to be an anomaly with the Dell machines.
So, how much is that in cold hard cash? We’ll look at some more surprising results in part 2 later this week.
I have just about had enough of the RM AppAgent lying to me about whether a system update is installed or not.
For some unknown reason, the RM package management system cannot tell the difference between a Windows update installing correctly, and one that starts to install and then fails halfway through. This means it will merrily tell me that all my workstations have Windows XP SP3 and Internet Explorer 8 when the reality is that a half dozen workstation have failed to install one or the other.
Why have they failed? Because the idiot who preinstalled CC3 on the workstation years ago relegated half(!) of a 20GB hard disk to the utterly useless local backup system, meaning I only have 10GB for the entire system partition – which has unsurprisingly run out of space, causing updates to fail installing.
This is the sort of amateur junk that CC3 is full of. Is it any wonder so many schools are reluctant to spend money on its overpriced and bug-ridden successor, CC4?
It is a travesty that in 2010 I still have to use a floppy disk to initiate a rebuild of most of my workstations. This frustration is compounded by the fact that on at least half of the occasions I need to rebuild a workstation, the floppy drive has long since packed up. Of those incidents, most are on machines that do not support USB floppy drives, so I have to swap out the internal drive.
CC3, I hate you more with every passing day.
I noticed early on at my current job that on one of my servers, I got the following message whenever I logged in:
This would be immediately followed by a similar and more lengthy message, this time regarding UDP instead of TCP.
Here’s what I found when I looked it up on RM’s knowledge library:
“This may occur if the server to which you are trying to connect has NetOp service enabled. You can safely ignore the error message. It is a normal behaviour when connecting remotely to a server via remote desktop.”
Translation: ‘We’re too lazy to fix this so it will annoy you every time you log in, forever. Love and kisses, RM.’
I got a quote recently from RM to upgrade the hard disks in one of my servers. This would involve buying three new 300GB SCSI disks and having an engineer fit them. I’ve been told I can’t fit them myself as that would invalidate the hardware warranty of the entire server.
The price I was quoted just for the drives was nearly double the price quoted by two competitors. I had anticipated them being a little pricier than normal, but this literally made me choke. In addition, the cost to fit the new drives was nearly £2,000, for a maximum of two days’ work.
I wish I could charge a thousand pounds a day, that’s for sure.
The total cost of the quote was roughly the same as buying an entire new server. I actually emailed the sales rep to point this out, and asked him if he’d made a mistake. He hadn’t. Personally I find it hard to understand where they think all this money is going to come from. The school has some very well-kept grounds, but as far as I know, there are no money trees on site.
When I started in my new job I resolved to keep an open mind about having RM as a primary supplier. So far, this approach is not paying dividends.
If you are the sales rep dealing with ADSL contract renewals, and I call you to ask why your product is so special that it is literally twice the price per year of your most expensive competitor’s equivalent product, then you should really start thinking about what makes yours different.
Telling me all the features that are standard on every ADSL product is a waste of your time, and mine. It’s worse if you actually can’t remember the correct terminology for the features and I have to finish your sentences for you after I get bored of waiting for the “um….” to finish.
However, the real killer for you would be failing to mention the one significant feature that actually makes your service unique.
I learned today that RM contract renewals are helpful and friendly. Unfortunately they don’t seem to have a firm grasp on their own product’s features, and some of those products are incredibly overpriced. Being helpful and friendly will get you far in sales, but it won’t let you allow your company to rip me off.