Chad V on 11 Feb 2009 06:12:39 -0800

[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]

Re: [PLUG] OT (but not really): Tough Interview questions

On Wed, Feb 11, 2009 at 00:37, Mag Gam <> wrote:
> Since the economy has crippled, I am certain Tech people have been
> feeling the crunch since most of us are considered costs rather than
> an assets. For those who have been looking for positions, I am curious
> what are some tough technical questions you were stumped at and for
> what position.
> For instance, I had a client as me over the telephone:
> What is ELF and how does it differ from other formats?  -- Unix related position
> What is kept in a journal of a journalized file system? -- Unix related position
> What is a poll and select? -- UNIX related position
> I consider these fairly tough questions -- what do you think?
> Just curious what are some other ridiculous questions you faced -- I
> am not talking about brain teasers or behavioral type questions...
> ___________________________________________________________________________
> Philadelphia Linux Users Group         --
> Announcements -
> General Discussion  --

I got sent this problem to see if I was even worthy to fly out for an
interview.  I've put about 10 hours into it so far and it will
probably take me another 10 hours to finish the script and write up
the reports.  It has been a pretty good challenge since the logs were
not in any standard format for web servers.



Online operations exercise / test:

 Supplied material:

    - 2 frontend logfiles from two different months of same machine
    - Format specification (Appendix A)
    - frontend server specs:
       - Maximum requests per Second: 120
       - Hardware acquisition time/duration: ~2 Weeks
       - Costs per machine: $2000

 Additional information:

    - We assume that the request growth is linear over time.


    - Senior Admin (on-site hardware deployment)
    - Junior Admin (supports on-site hardware deployment)


    - Offshore data center (located in another city)

 Problem description:

    - Get the relevant key data from the logs that show the current
frontend capacity utilisation.
    - Document the process / tool chain or your self made scripts used
to extract the key data.
    - Analyze when the next front end machine is required by
projection of key data on a timeline.
    - Create a decision supporting document that makes clear why and
when hardware acquisition is required.
    - Make a draft project plan and schedule for hardware acquisition
and deploy.
    - Isolate and document the tasks that needs a follow up in the
deploy process.


    - Any tool, language and source of information is allowed.
    - If you developed scripts to extract the key data please supply
them with the results/material.
    - Additional thoughts that would lead to a better prediction are welcome.
    - Even though the data is anonymized suppose that it is
confidential. Handle with care.
    - Stay cool.

Appendix A)

 Accesslog format definition:

Field seperator is : \t

Field Index:

 1. log-level: (INFO|ERROR)
 2. time stamp - (readjusted to match simulated case)
 3. request-id
 4. total request processing time
 5. 90% percentile total request processing time
 6. status code/status message
 7. thread-id
 8. request URI - (anonymized in the supplied logfiles)
 9. referer - (anonymized in the supplied logfiles)
10. remote address - (anonymized in the supplued logfiles)
11. session id - (anonymized in the supplied logfiles)
12. user id - (anonymized in the supplied logfiles)
13. common name / user name - (anonymized in the supplied logfiles)
14. exception
15. backend request time
16. sum of backend request times
17. 90% percentile of be request time sum
18. session db request time
19. sum of session db request times
20. 90% percentile of session db request time sum
21. acc db (hdb) request time
22. sum of acc db (hdb) request times
23. 90% percentile of accdm request time sum.
24. list of db request times
25. sum of misc driver requests
26. sum of db request times
27. parameter map (anonymized in the supplied logfiles)
Philadelphia Linux Users Group         --
Announcements -
General Discussion  --