Customers have asked about our maintenance schedules before, so I’ve put together this post to give you all an understanding of how we work and why and also answer some of the most common questions.
When do you perform your maintenance?
We perform this overnight on the last Wednesday of each month. We have a set period as this means we can always make sure we have the correct staffing present to make sure any maintenance goes as smoothly as possible.
Any different tasks that crop up requiring non-urgent maintenance will be added to this monthly schedule.
What sort of tasks do you perform?
Generally, it’s housekeeping. We need to make sure our platforms are always up-to-date with any new patches and releases and optimised to run as efficiently as possible.
Keeping on top of all this means we can schedule in some planned maintenance (often with no noticeable effects to customers) and ensure we do not ever risk a situation occurring where unmaintained servers start suffering from performance issues, or a worse case scenario of unscheduled downtime.
Will the maintenance always involve downtime?
No, the majority of the maintenance will not be noticeable to the majority of customers.
Our shared hosting is all load balanced and run in large clusters, so our sys-admins can work on one particular cluster whilst the others carry on serving as normal. The days of single-points of failures are generally in the past for most shared platforms.
Some work to specific VPS hosts or certain network aspects may involve unavailability, however we always keep this to a minimum.
There has been maintenance on dates other than the last Wednesday of each month. Why’s that?
Basically, we need to make a judgement call when a new issue develops.
If something is fairly minor and will not have any negative effects, it’s fine waiting until the monthly window to fit into the normal schedule.
If something poses a risk, however, we have to decide if we need an interim maintenance window rather than waiting until our next scheduled maintenance window.
The classic example here is new security vulnerabilities in operating systems. When these are announced then they really need addressing straight away or we would be putting customers at risk. It would be irresponsible of us to drag our heels with something like this if the maintenance window was not until a couple of weeks away, as that’s a couple of weeks we would be leaving our customers vulnerable.
I hear other providers don’t have as many maintenance windows as Daily. Why is that?
We can’t speak for other companies specifically (firstly because I don’t know their internal policies and secondly because I don’t want to get into trouble!), but the example I often give to customers who ask is this:
- How often does your home PC have a new Windows Update?
Forget about service packs, let’s just say security updates for windows, networking, IE, bug fixes etc. There’s probably a few a month at a guess. It’s just the same on servers in this respect (if not more so).
Now if there’s a new security update for, let’s say IIS 7 on Windows VPS, then that’s going to be the same for IIS7 at Daily.co.uk or any other company running that system.
If you think that it’s bad that Daily are scheduling in maintenance to address a known security vulnerability and it’s good that another company are leaving a known vulnerability active on their systems, then just play that through in your head for a few minutes and think of the implications. Is it really that good that for the sake of avoiding a planned maintenance window at times designed to minimise disruption, your site and data is at risk from exploit?
Let’s put it another way: If the brake warning light on your car dashboard comes on, do you get your brakes checked or do you just take the bulb out of the warning light? One scenario means you have to get your car into a garage and then carry on as normal; the other scenario means you’ll probably carry on fine with the “out of sight, out of mind” attitude until you find yourself whizzing across a roundabout with no brakes looking like you’re in the Batmobile.
The fact that you’re not being told about a problem does not mean there is not a problem. Any sys-admin who sticks his or her head in the sand and hopes a problem goes away is turning their network into a hacker’s paradise!
So do not all providers keep their servers and systems updated?
Again, we couldn’t comment on specifics, but if in doubt – ask them.
Most companies should be able to list any of their recent updates and the reasons for them and you can then see for yourself what sort of durations you are looking at.
If you have a VPS, then also check to see if they do even perform updates themselves. Hopefully they do on the hosts and hardware, but you may be required to manage any updates yourself within your VPS.
So does Daily perform maintenance too frequent, too slow or just right?
It’s not an exact science, but to summarise:
- We schedule a monthly maintenance window in which any non-urgent tasks can wait to be carried out at once.
- If anything urgent comes up in the meantime, we’ll perform high priority maintenance if there is a valid security risk or if performance could be decreased significantly by leaving it until the monthly window
- We also make sure that as much maintenance as possible is seamless. Downtime is always a last resort
When it comes to risk taking, everyone has different ideas of what’s acceptable but we always act as is befitting of a responsible host.
If anyone finds any maintenance an inconvenience, we do apologise – that’s not our intention. However we’d like to think you can sleep sounder at night knowing any outages are always as short as possible in periods you know about in advance, rather than trying to access your services in the middle of the day and finding them off-line!