Contribution from PaulS written on 4/28/2004 at 02:52
Below is an overview
of my approach to CNI qualification, comments and feedback are welcomed.
Assume that the DQ has been done and the URS is defined and the network
has been supplied to meet the requirements.
Imagine that we have just built our network (even if you are
retrospectively qualifying your CNI) and have installed OSs and the
Client/Server s/w. No applications exist and are outside the scope of the
CNI qualification. The components that are in the scope have been defined.
We get topology drawings and actual cable run drawings. We make sure that
all the components are installed as per the manufactures recommendations
(including matching the hardware specification for software i.e. Enough
memory etc) or as standards requires (cable run lengths, EMI/RFI
interference). We capture the configuration details of the hardware and
software between the end points:
e.g. Client s/w-OS-BIOS/CMOS- network adaptor card-transport
protocols-network adaptor card-BIOS/CMOS-Server OS s/w.
This includes DNS, WINS, DHCP, WINSOCK, RAS, TCP/IP, etc. The above is
considered as the IQ.
We perform basic operational testing, boot up, shut down (of
Clients/Servers) client logon/log off, access permission's to folders can
be configured and some dummy data can be copied-pasted-moved-deleted etc
between selected locations. Printing is confirmed as WYSIWYG. The above is
considered as the OQ.
We perform PQ tests (based upon the URS) where performance is monitored,
audit trails are checked for completeness, time date synchronisation
confirmed etc. A baseline of network performance is established over time
(typical week or month) and will be maintained, as defined in the Service
Level Agreement with the IT department.
SOPs are written that define change control/disaster
recovery/backup/maintenance repairs/workstation testing etc.
I envisage the workstation testing SOP will state that workstations, with
their various applications will be tested (start up, shut down, network
communications and major functionality) on a sort of network test bed or
segregated network segment if possible. Manage the stored image of
workstations (use groups of standard builds and configurations) and
install new applications on an image replica, test for application
conflicts and alike prior to being let lose on the live network.
SysAdmin work done on the configuration of the parameters between the end
points, as stated above, gets most attention from change control, to allow
room to manoeuvre for day to day SysAdmin tasks that will not have a major
effect on data integrity. BIOS/CMOS settings are related to time/date and
sleep modes (of clients) that can cause links to drop and lose or
corruption of data.
Have I missed anything major? Done anything unnecessary?
Answering this Contribution: