Well? What was the first thing you thought of? Bankrupt? Stupid? Behind the times? Surely this must be a trick question. It’s 2016. Every business has a website. Right?
OK, it was a trick question. While more than half of small business do have an actual website — temporarily ignoring the fact that that means there are still about half of small business that don’t have a website — the majority of them aren’t doing enough to proactively make sure it’s working. You shelled out the chunk of change to choose a web design company in the first place, so why wouldn’t you get the insurance to protect it?
Because of limited budgets, companies today are being more careful about where their IT dollars go to, and as a small business, you may not have the resources to devote to a full-time in-house monitoring system. And frankly, when your website is working right now, it’s not high on your to do list.
“Effective network monitoring is much like a disaster recover plan,” said David Nizen, vice president of business development at iGLASS Networks, a network monitoring service provider for small businesses and larger corporations. “You only need and appreciate it when a disaster strikes, but if you keep putting it off, you’ll eventually get burned. Often that disaster will turn out to be way more expensive than it would have been if you planned ahead. An ounce of prevention is worth a pound of cure. (Ben Franklin said that.)”
How bad can it really be?
You know how imperative your online presence is. It’s the heart of your overall marketing initiatives and a significant source of leads and revenue for your business. And supporting all of that is your business’ core IT network. Your network is the foundation upon which your business is built, powering and supporting your website, email, point of sales and so much more. It connects your customers to your services and goods, employees to sales and support systems they need, and management to the data and applications they need to make critical decisions. So what if it in the middle of the night, while you are slaying dragons in your sleep, your network drops? How bad could that be?
Loss of employee productivity. Loss of customers. Loss of revenue, about $1.7 billion each year according to CDW survey. Customers ill-will from a bad experience with your business. Loss of business opportunities to competitors who are just a click away… The list goes on. In short, pretty bad.
“Some of the worst things that could happen are outages which idle employees or prevent customers from doing business with you,” Nizen said. “The most painful ones, from your boss’ perspective especially, are outages which could have been prevented, had you only known about the problem ahead of time.”
What problems could I see?
While network problems come in all flavors and sizes, Nizen identified some of the most common issues that could happen:
- Full disk drives
- Systems running out of memory
- Too high system load
- Key applications or processes dying
- Key websites failing to load
- Physical hardware failures
- Network outages (fiber cuts, rooting loops, touring snarls)
- Network latency
Still, the worst problem that could happen is one that could have been prevented (or resolved faster). The runner up is just not knowing it’s happening at all.
What can I do to prevent them?
The good news is that there are a few precautions that small business owners can take to ensure that network downtime is at the utmost minimum. The SLAC National Accelerator Library at Stanford as a running list of network monitoring tools, but it boils down to four options:
- Hire an in-house IT guy: This will get expensive as tech support is not cheap, and you have to hope that he does not need to sleep in the 24 hours a day, 365 days year that you need your network to work.
- Free software: Yes, the free word is inciting, but don’t call it a day yet. Just like stock photography, you get what you pay for, and you or someone on your staff will still need to get up to speed on deploying and maintaining which ever package you decide to gamble on.
- Paid software: You get a more robust option and can customize it to fit exactly what you need, but you’ll still become your own IT guy trusting that a computer-based program will maintain a computer-based network program will skillfully maintain and monitor a dynamic, changing IT network.
- Outsourcing: It’ll still be pricey, but it’s the most cost effective option that gives you full protection since you’ll have engineers doing the monitoring while you’re running the business. You also won’t have the headaches associated with training, vacations, sick days and attrition
What can I do to keep it all in check?
- Don’t ignore syslog and traps: Many monitoring platforms do a good job of polling using protocols such as ICMP ping and SNMP, but many folks fail to invest the time to set up syslog messages and SNMP traps. This kind of telemetry can be invaluable in catching errors and critical issues that are reported by key components such as servers and routers. You’ll need a trap collector to process and alert on this data, but it is well worth the effort.
- Important data deserves thresholds: Too many times customers deploy monitoring systems that collect tons of data, but don’t have any thresholds specified. It’s great that you can look at pretty disk utilization graphs, but if you don’t put a threshold on it, you’ll never get that alert at 3:00 AM telling you you’re about to run out of space.
- SNMP isn’t always the best method: SNMP is usually the workhorse providing the majority of our data collection. That said, some equipment either doesn’t support it or had done a poor job implementing it. Often times there is a call level interface (CLI) or other protocol more suited to collect the data you need in order to do the job right. For Windows servers especially, use WMI/WinRM for data collection and a utility like SNARE to forward Windows Events as syslog messages. The platform you choose needs to be flexible enough to incorporate multiple methods of data collection.
- Don’t rely on agents: Agent-based systems require you to install a small piece of code, or program, on the machine to be monitored. If there is a problem detected on that machine, the agent sends an alert to the monitoring system, which then typically sends an email or SMS to you. The problem with agents is that if the machine fails and goes down, so does the agent. Use a system which is independent of the network it intends to monitor.
- Consider using synthetic transactions: How many times have you gone to a website just to find out that you couldn’t log in, post, download, etc. Just because a site is up and accessible doesn’t mean it’s actually working. Synthetic transactions allow you to do a ‘deeper dive’ and actually put a site or application through its paces. The automatically, and systematically, test for availability and correct functioning of key systems. For example, a synthetic transaction may simulate accessing a website, logging in to the site, downloading a file, then logging out – every 5 minutes.
- Have a backup plan: To guard against disaster, you need backups. Whether it’s the configs on your routers or the files on your PCs and servers, there are numerous services out there to help you get back on your feet in the event of a catastrophic failure. If you’ve ever had a disk drive fail on a PC or laptop, this one should be obvious to you. Companies like Carbonite and Mozy make inexpensive backup services available to the masses for as little as $50 a year for consumers and just a little bit more for small businesses.
- Ease of working together: That foreign-based company may have some killer pricing, and even an effective monitoring platform, but when they call you at 2 AM will you be able to understand them?
Proper monitoring can be tricky to implement and it does require a great deal of time, effort, and skill to maintain it. There is hardware and software to buy, analysis and development to be done, implementation and customization, and then the ongoing updates, upgrades and maintenance. But don’t wait until you’ve gone through an outage once to take the steps to put some sort of system in place.