Make sure you order support with your dedicated server
It's an unfortunately too common situation that we come across a customer in a very distressed state, sleep deprived, anxious and often venturing into the verbally abusive mode. The source of all this heartache - their web application service has been offline for minutes, hours or even days and their customers are hammering them.
Now don't get me wrong, this is not an admission of serious negligence, bad support or systemic problems at Anchor in any shape or form - if we had such problems I don't think I'd be writing about them.
The unhappy support calls tend to play out something like this:
- Customer: My service has been offline for x days, it must be your fault
- Anchor: Well sir, this is the first time you've logged a fault.
- Customer: I don't care, this must be your fault. You provide the server, you should be monitoring it.
- Anchor: Sir, the server plan you're on is the self managed one, you chose the one where this is your responsibility
- Customer: Can't we go to backups, won't that fix it.
- Anchor: But Sir, when we sold you the server you decided that backups were too expensive
- Customer: But surely you do backups anyway right?
- Anchor: No, we don't. Would you like one of our systems administrators to look at this and resolve the problem? This is a consulting service so there will be a cost involved.
- Customer: What the beep beep beep. My server is broken and you want to charge me more money? I can't afford that. Do you know that I have Mutinational corporation XYZ on this server and there are now hundreds of people that can't do any work because this is down?
Yes - we really do get calls that play out just like this.
At this point, we're scratching our heads. For people that deal with systems administration on high availability services day in day out it seems only normal that if you have a very high level of dependence on a website or any other service for that matter, there needs to be a commensurate level of expenditure on monitoring, maintenance and planning to prevent faults and allow rapid resolution when problems occur.
This article with support side of the dedicated service as I believe that this is the most common problem area we see. The other components are discussed in our dedicated server purchasing guide.
Abstractly speaking you could argue that the reason our theoretical customer found themselves in this dilemma is for one of two reasons:
- They were was sold the wrong level of support - ie, based on what we knew about their requirements we sold them the wrong support plan, or
- The customer having been made aware of the support options was not prepared to purchase a suitable level of support.
Unlike other companies, we're pretty up front about our support plans. They're detailed and priced on our website. I like to think that we do a reasonable job of making the options known to the customer. Let's examine point 2 first by considering some of the reasons that the customer may not have been prepared to purchase the correct solution:
- They were given a range of support options but they didn't understand them.
- They thought that they were being sold something that they didn't really need and avoided it. Kind of like upsizing your Big Mac meal.
- They didn't understand the different parts of the hosting service
- Budget constraints
- A startup business that just doesn't have the funds
- A suitable budget hasn't been allocated
- Spending less money on hosting means better company profits.
- They didn't appreciate the risk associated with not spending money on support.
I believe that if we address that last point - appreciation of the risk associated with not having the right level of support - we can solve the entire problem. Our experience in the industry over the last 8 years tells us that customers who do have this risk appreciation don't baulk at the cost of support. An appreciation of the risk implies an understanding of what they're buying. It means they know that this is important.
How do we get customers to understand the risks? We write articles like this and hope that they will read them.
We asked ourselves the question recently, why is it so hard to get some customers to appreciate the risks of not having a sensible support arrangement in place? The overwhelming conclusion was that what we do is largely invisible. We're asking people to pay for something that's intangible. For all they know, this could be a case of the Emperor's new clothes.
My background is engineering so I love analogies, diagrams and pulling things apart. I'll spare you the diagrams but we'll work on some analogies for now.
Take a car. It's your weekend driver. You mainly use it to go out on social occasions and do the odd bit of shopping. You're environmentally conscious so you take the train to work each day. It's a new car, nothing fancy but it does the job. You get it serviced once a year and aside from a flat tyre once in a blue moon it never lets you down. Your level of dependence - low.
Let's compare your daily driver to your home PC. You use it to check your email, read the paper online (to save the trees), do the odd bit of facebook, some cheap phone calls on skype to friends overseas; that's when you've managed to wrestle the kids off it who've filled it with viruses whilst looking at things they probably shouldn't be. But it works. If it falls over, well, the kids go outside and play and you use your mobile for the phone calls. You take it down to the computer shop on your way to work a week later. It's fixed a week after that and you've enjoyed the two week computer absence and picked up a book whilst the kids got some exercise. Your level of dependence - low.
So all cars and all computers are the same, right? WRONG!
Take a Formula One race team. The only justification for the hundreds of highly paid people, highly stressed jobs, working through the night, jetsetting around the world, burning through millions and millions of dollars are year is to make sure that just one car, yes this is all for one car, is performing at 100% on race day for a race that lasts an hour or so. The engine might only last that single race, but that's ok. The level of dependence - high.
The web server that runs your online application. It needs to be running 24 x 7. You measure the amount of time it's working as percentages above 99.8% each month. Your business stops selling if it's not working. Your clients don't get what they're paying you for if it's not working and they in turn have to stop working. You're constantly tweaking the server, making changes, adding new features but all of these changes must not disrupt delivery of services. The load on your server is constantly changing. If it breaks at 3am in the morning, it can't wait till the next day to be fixed. Your level of dependence - high.
Yes, your home PC might run Windows and your server might run Windows, but the things you need to do to look after them both are very, very different. The whiz kids that we have looking after your machines didn't spend 4 years at university playing games or hacking - they picked up the skills which lets us achieve uptimes of 99.8%.
By the time you're on your own server, whether your wallet believes you or not, your level of dependence on your server is high. What I hope is quite clear by now is that if you have a high level of dependence on something you need to spend the money to get a level of support that will deliver a level of service that is inline with your expectations. If you skimp on the support, or don't find a provider that can deliver the right level of support - chances are you'll be enjoying one of those conversations above sometime soon.
Some food for thought:
Did you know
- That we see on average over 100 attempts to break into every server every day.
- We've seen servers with no publicly advertised services hacked within 3 hours of going live due to easily-guessable passwords.
- That hundreds of security holes are found in common web applications every month that could let someone hack your server.
- Most hacking attempts are automated - you don't have to be high profile to get targeted.
Have you thought about:
- What happens when changes are made with no change management plan and there are unintended consequences?
- What would happen to your business if a fault took your web service out for an entire business day? Or longer?
- How much outages cost you both in terms of direct loss of sales and damage to your reputation?
Do you have:
- A single IT person looking after support on your server? Are they on call 24 x 7? What happens when they're sleeping? In the pub? Partying? At the footy? On holidays?
- A plan for dealing with outages on your web service?
- Monitoring of applications in place so that you know if there is a problem?
- Monitoring of the load on your hardware? What happens if you run out of disc space?
- A plan for applying updates to applications to maintain security?
- A firewall in place?
A policy for ensuring only strong passwords are used?
- A preventative maintenance plan in place to stop problems occurring in the first place?