Bladewatch.com

I was having a chat with a Beatrice, she’s started work at a smaller business in the North of England, she reminded me about an important statement for all those new and existing server owners/administrators, I’ll let her email speak for itself, I’ve removed any references to the company involved:

I’ve been here a few weeks and I keep coming across servers that have been up for over a year or so, that ‘work’ but as soon as you try to log in to them, or do anything they seize and end up needing rebooted. Avoiding the conversation about security patching, what’s your view on server uptime and rebooting?

This morning we had to reboot our web server as it got slow and was out of memory, so the web site was down for about 10 minutes because we hadn’t rebooted it for nearly a year out of hours.

Regards

Beatrice

If you speak to a Unix or Linux guy they tend to love to mention the lack of reboots, the industrial strength of their operating system, the number of actions that might require a reboot in Windows and in some respects they’re right, but at the same time just ask them how many servers they have that no one can remember rebooting and are therefore too scared to do so in case it doesn’t come back.

Having abstracted ourselves from the Windows vs Unix conversation, I’ve always thought a server should be rebooted every few months, to clear down the memory, to maintain service and to apply drivers/firmware/layered component updates. Can we leave a Windows server online for years, absolutely, but we’d be missing security patches, and we run the risk that the server becomes unhelpful at the point at which you need to use it. We need to change the mindset from olden days of how long has the server been up for, too one about scheduled downtime windows during which IT can action best practice support and maintenance, (planned or unplanned), it’s only by maintaining the systems properly that we get the best efficiency and delivery from them, and as I like to point out to users/colleagues we can either reboot it on Sunday when it’s not in use, or mid Monday morning during trading because there’s an issue.

With the Microsoft security patches being released every month, there in theory shouldn’t be servers that are up for more than a few months, this is good for IT and the end user community, granted there are 24/7 businesses, 24/7 systems, but regardless of the platform, we need a window to do much needed maintenance to keep the system operating as expected, to prevent the world ending scenario “… apologies, our computers are down can you call back to order later?”.

No related posts.

Related posts brought to you by Yet Another Related Posts Plugin.