rb|it : blog

Is uptime the best measure of system efficiency?

One of the main responsibilities of IT staff is to keep the company systems up and running, the success of which is commonly measured as percentage ‘uptime’.  For the business uptime = money made and downtime = money lost; the aim is to get the uptime figure as close to 100% as possible.  That figure is most commonly based on the response to tests, by network monitoring systems, on one or more specific servers or services.  The weakness of this approach is that a network might have a server uptime of 99.99%, but the average uptime on a company PC may only be 85%.  It's also possible to fall into the trap of obsessing about uptime when it's not as important as you might think.  I take a more pragmatic approach, that downtime is only an issue if it effects the users of the system.

For most companies, it's more productive to expand the definition of ‘downtime’ to include ANY reduction in functionality on ANY system during operating hours.  Now a truer picture of uptime appears (though the 100% target becomes even more impossible to reach).  This is the approach I prefer to take, focusing a little wider on the organisation as a whole, rather than exclusively on the servers and their uptime.

Unfortunately it’s very difficult to measure uptime in such a holistic way, based on every server, switch, router, PC, printer or mobile device.  Traditional network monitoring systems fall short of the task as the complexity of the functionality they would need to test is too great.  Only a person can make an assessment of whether a system is 'up' or 'down'.  For example, the PC may be up, Outlook may be running, and the mail server may be fine; but there is a permission error on a public folder so the user can't see what they need to see.  For the user, THAT is downtime.

An approach to get a better understanding of your total system performance is to use your helpdesk or other issue tracking system to track reported incidents and their impact.  If you educate and encourage your users to record every issue they encounter, and you record how long it takes to resolve the problem, then you have a better picture of the real uptime of your network.  That picture allows you to focus your resources on ALL causes of downtime, rather than the traditional priorities (server or component failure, power outages etc.)

Taking this approach will identify the causes of downtime which are often missed; poorly configured client software, incorrect permissions, user training etc.  Focus on these issues, and you will increase the real uptime of your systems and ultimately the productivity of staff.  It's not as conspicuous or immediately gratifying as a 99.99% uptime figure, but it's ultimately more meaningful!

The good, the bad and the ugly

More of a news round up this week than a monologue, three things that have caught my eye over the last seven days.

The Good
Earlier this month HP launched a babysitting service for SMB's called 'Insight Remote Support'. It's available to anyone with valid warranty or Carepaq on a Proliant Server (bar the ultra low end 100 series), BladeSystem chassis and Proliant Blade Servers, and the StorageWorks EVA midrange disk arrays.  All you need to do is install the agent from the HP site and they will monitor your server for you, making sure it all works, warning you of impending failure, and even ordering spare parts to replace faulty components and dispatching them with an engineer to fit them.  When 90% of of HP's SMB customer base is worried about unplanned downtime it will be a shot in the arm and potentially a huge money saver.  For more info see http://h18006.www1.hp.com/products/servers/management/insight-remote-support/overview.html

The Bad
In October last year I blogged about using the 'inurl' switch in Google to exclude unwanted domains from your search results.  Having recently reviewed this I found some interesting things I either missed last time, or have changed.  Firstly, the Google page on Advanced Search Operators does not state that you can use the inurl switch with the minus operand to exclude words from a URL.  Secondly, there is apparently a restriction on how many words you can enter into a query (32 words), which restricts the amount of domains you can exclude using this switch.  Lastly, I have seen in some search results that addition of an inurl switch actually increased the result count!  Whether this is an attempt to discourage use of this switch I can't say, but it certainly isn't encouraging.  The switch still appears to have the desired result though.

The Ugly
A new(ish) development in network security is the compromise not of home PC's, but home routers.  A worm called "psyb0t" has been detected in the wild by a research group DroneBL, and is estimated to have infected at least 80,000 hosts.  Until now, 'botnets' have been made up of PC's with broadband connections that are infected with a virus or worm, the weakness of this method being that the malware is potentially detectable on the PC by the user.  However, compromising the router (which the user knows almost nothing about) is a far more stealthy method of infection.  This worm appears to have been deactivated by it's author, and possibly the result of a research exercise (though still illegal).  However, now the concept has proven to be successful it no doubt adds another weapon to the arsenal of the bot herder.  I would strongly advise all of those who can to check that their DSL or cable router has remote management (whether that be web interface, ssh or telnet) disabled on the internet interface.  If you must allow remote mangement, at least restrict those IP's that can access it, and use strong passwords.

Open Source Issue Tracking

Customer Retention
Almost any company providing a product or service receives issues or problems from their customers.  They might come in by email, by phone, or even by post, but those issues need to be addressed, and the process for doing that can be problematic.  In many smaller companies there is no software system (or even process) for dealing with these issues, and customer satisfaction suffers as a result.  Several articles recently have identified customer retention as a key survival tool in a recession.  Those who provide good customer service will be seen as ‘high value’ providers, and those who fail to deal with customer issues will lose customers to their competitors.

You get what you pay for
In the past, ‘Helpdesk’ or ‘Issue Tracking’ systems have been very much a case of ‘you get what you pay for’.  The lower priced systems were simple at best and lacked critical functionality (like automated issue logging from email), and the better systems were expensive, often beyond the budget of SME’s.  There is an alternative though, and it looks very promising.  rb|it-Consulting has recently been evaluating the latest version of OTRS (short for ‘Open-source Ticket Request System’), and is now adopting it as a Helpdesk and Issue Tracking System.

The good news
The software is entirely open source and free, runs on Microsoft Windows and UNIX/Linux and comes complete with all the required packages.  The installation is simple and works on both Windows Server and Small Business Server 2003 (the latter requires a tweak to the installation to avoid clashing with IIS).  The end result is a web based system which works in all browsers (including mobile devices), which can be integrated with your chosen email system (whatever that may be), and provides an effective and powerful tool for tracking issues from receipt to conclusion.

Features

1. Automated incident logging via email
2. Automatic email notification to customers and staff of changes to issue
3. Full function web based console for agents and customers
4. Built-in escalation of issues and other time based functions
5. Calendar support (to define working hours and days)
6. Service and SLA Support
7. Extensively customisable via the web interface

The default installation would require minimal configuration for a simple setup, but the benefits are potentially huge.  Your customers would immediately know that their issue had been received and have a unique reference for their problem, and can log into a web interface to check or update their issue.  Your staff can see all the outstanding issues, prioritise them, see which were overdue a response and can assign them to whoever in the organisation (or outside it) is responsible for fixing it.  Take a look at the demo version (see http://www.otrs.org/demo/ for details) and try it out for yourself, I think you’ll find a free, powerful and easy to use system which will make your customers happy, and your Financial Director even happier.

Richard Bartlett offers IT consultancy services through his company rb|it-Consulting.  Contact him by email richard@rbit-consulting.com or visit his website at http://www.rbit-consulting.com

IT Support: money for old rope?

Paying for IT Support is like paying for garage services, or an electrician.  It's magic, you don't understand it, and there is the potential for you to be charged for things you didn't need.  So how do you make sure you're getting value for money? 

Buying IT Support

1. Stabilise your network.  Before investing in an IT Support service, get your network stabilised.  A ‘one off’ network audit/health check and resolution of all current issues will pay back within a year.
2. Measure your demand.  Monitor how many technical problems you have (whether user problems or actual system failures) over a fixed period of at least a month.  Decide how important each system is, and what the impact of a system failure would be.
3. Fix your budget.  At the moment this will be a ball park, and defined mostly by existing expenditure or cash flow restrictions.
4. Tender.  Clearly state what service you are looking for, the number and type of systems and the necessary response/fix times as appropriate.  Use your contacts to get recommendations, and Google to identify potential suppliers.  Ask for and talk to reference customers, and check the contract thoroughly to make sure the service is clearly defined.
5. Monitor.  Make sure any ‘SLA’ or response time  in the contract is adhered to.  Talk to your staff to see if they feel the supplier is responsive and helpful.  If the company doesn’t perform, use the contract as leverage to make sure they do; and don’t be afraid to drop a poorly performing supplier.

Caveats

1. Avoid up-selling.  Ideally, separate your IT Support supplier and IT Hardware/Software supplier to avoid the potential for ‘up-selling’ you hardware on the back of a system failure.
2. Buy hardware with a manufacturers 3 year warranty, fixing your hardware support costs for three years.  All you should pay for now is small labour charges per fault.
3. Only pay once.  If you are paying a charge per call (or per visit), and a problem recurs, say when you report it the second time that you’ve already paid for this to be fixed once, and you don’t expect to pay again unless you caused it to recur.
4. Charges for anti-virus definition and patch updates.  Some things shouldn’t be charged for, and anti-virus definition updates and Microsoft patches are two examples.  Both are built into the respective software, and designed to run without intervention, so don’t pay a company to do this for you every month.

Follow these steps, you should find your IT Support provider is a real asset to your company, rather than a drain on your budget!

Is Software as a Service the right choice for you?

In the search for better value solutions, businesses often come across SaaS (Software As A Service), where software is provided and managed from a central point on demand, rather than installed on local systems and managed locally.  An example of SaaS is the hosted Microsoft Exchange and Sharepoint systems which now provide an alternative to in house email and intranet systems.

The value propositions of SaaS include;
(a) it's priced on a per user basis, with no 'up front' capital investment
(b) the minimum number of users licenses is very low
(c) it's total cost of ownership is low as maintenance and backup costs are incurred by the service provider

This model can put quite advanced IT systems in the reach of small businesses, who otherwise could not justify or support the expense.   For example, comparing the costs of hosted Exchange and Sharepoint to Small Business Server, it's clearly cheaper for a startup with 3-4 staff to adopt the hosted option.

Is it cheaper for all businesses in the long term though?
rb|it-Consulting recently did some analysis on the cost of different IT options for SME's over a 3 year projection, and found that the cross over point (where hosted Exchange and Sharepoint became more expensive than Small Business Server) was somewhere between 5 and 15 users, depending on the cost of technical support.  Certainly for a company with more than 20 employees the pricing models used in my analysis showed it was no longer cost effective.

An article by Gartner on Monday echoes this, stating that "although software-as-a-service (SaaS) is cheaper in the short term due to the lower capital investment, total cost of ownership could turn out to be more expensive in the long run".

Whether the long term cost of SaaS outweighs the advantages is still worth investigating, and the upsides (lower capital costs, reduced total cost of ownership, faster adoption) cannot be ignored.  It is worth remembering though that if you are considering an SaaS solution, the only way to compare it to traditional in house solutions is across a time line of at least 3 years.  It is also worth remembering some of the downsides of SaaS, which include;
(a) data is transferred at internet speeds rather than local area network speeds
(b) a higher cost of storage per GB compared to local storage
(c) reduced ability to customise and adapt the solution

Having said all this, for startups with up to 5 staff, it's difficult to argue with the value proposition, and gives those companies an easy way to adopt valuable IT systems in that critical first year without a big up front capital expense.