Lecture no 24 Helpdesk

1
Lecture no 24 - Helpdesk
TDT4285 Planlegging og drift av
IT-systemer Spring 2007 Anders Christensen, IDI
2
The role of the Helpdesk
Single point of entry
IT-sysadmin- staff
Customers
Helpdesk
E-mail Physical access Telephone Web Chat
Porjects
Interrupt- controlled
3
Single point of entry why

• To maintain control over the speed
• To make things more uniform for the users
• To prioritize between different tasks
• To get an overview over which tasks remains
• To limit interruptions to a small part of the
  organization
• Centralize expertize on user communication

4
Key points for helpdesk

• Not everybody is suitable for work there
• Consider interior and location
• Estimate sufficient staffing
• Announce it actively, target user groups
• Monitor it for performance
• Use relevant tools to support the helpdesk

5
Suitable (?) personalities

• If we stopped caring about the customers,
  maybe they would stop bothering us?

Demotivational poster
6
Location and interior (if physical)

• Location
• An area where lots of people pass by
• Preferably a neutral area
• Interior
• Sufficient room (space and air)
• Friendly colors and furniture
• No hang-arounds, old friends or pentioneers

7
How much staff is sufficient?

• Ratio between helpdesk staff and users depends on
  the situation
• University environment (emploees) 50-100
• Students 300-2000
• ISP 5000-50000
• Note these numbers are only examples (and they
  change all the time)

8
Information and announcements
Location
Focus
Nobody read the documentation
Timing
Briefness
Docs must be Especially designed
Repetition
Variation
Relevance Usefulness
Comprehensible
9
Metrics for helpdesk

• Number of customers by helpdesk staffer
• Volume of contact, and the change over time to
  this volume
• Degree of fulfillment of SLA
• Need for personalized training and education
• Share of indcidents that is escalated

10
Metrics for timing
Problem
Closure
Symptom
Final feedback
Fixing
Error message
1 minute
1 hour
First feedback
Note the times given is just examples for
discussion
Analysis
10 minutes
11
Incidents in a helpdesk

• Priority
• Ownership
• Contact-info and customer info
• Timestamps
• Refs to CIs
• Status
• Categorizations
• History

• Connect similar cases
• Close old cases
• Dispatch cases
• Make statistics
• Search in old cases
• Prioritize between cases
• Escalate cases

12
Priority and parameters

• Coverage How many are covered by the error (1
  5 100 all)?
• Severity what is the consequence if the error
  is not corrected (beauty-error inconvenience
  error crisis)?
• Frequency how often does this error ocurr?
  (continually daily monthly once)?
• Status how importen is the reporter of the
  error (your boss -- paying non-paying)

13
State diagram for an incident
New
Analysed
Archived
Open
In work
Closed
More info
Wait
Note different tools have different paradigms
14
Definition of the coverage

• Which machines and applications
• Who can contact the helpdesk
• Where can they get help
• When can they get help
• How long must they expect to wait
• How is the incident escalated

15
Tools

• Trend analysis to predict the future.
• Statistics to say something about a large
  number of users and machines
• Qualitative feedback what does the most visible
  users think?

16
Helpdesk phases
Problem identification
Planning and execution
Welcome
Verification
Greet them
Classification Description Verification
Alternatives Selection a Solution Implementation
By sysadmin By user Closure
17
The Welcome phase

• This phase is often under-estimated (But, it is
  just a nice hello)
• It is often a critical phase it is a first
  impression, and some users can easily feel
  dismissed.
• If done well, it will facilitate the next message
  from that user.

18
Problem identification

• Classification which area is this prolbem
  within. Might be partly automatic
• Description Make a short and explicit
  description of the incident,
• Verification check that the problem is
  reproducable

19
Execution phase

• Suggest different solutions maybe on temporary
  for today and a possible permanent for the
  future. Often a task for experts
• Select a solution prioritize between the
  solutions, involve the user, and a suitable
  solution.
• Execution let a sysadmin execute the solution
  if neccessary, or otherwise help the user

20
Verification phase

• Sysadm-verification Let those who have changed
  something or suggested a solution test it before
  forwarding it to the user
• User-verification Let the user himself test the
  solution, and come with feedback on whether the
  solution worked or not.

21
Incorrect roles

• The bogeyman frightens the user away no work
• The forwarder sends the user somewhere else
• The assumptionist knows what the problem is,
  even without having studied the problem and
  symptoms
• The conceptualist who fixes minor things by
  changing everything
• The typo-ist who never is able to write docs,
  commands, or scripts without typos.
• Hit-and-run-Sysadm comes, writes commands, run
  (dont tests)
• The Sandman closes ticket