2007/01/10

A new world of IT - appliances, clients, infrastructure, services

A thought on some new taxonomies in the I.T. world.

Even the meaning of "I.T." has changed - it's now Business I.T.
Somehow that also includes what we do at home...
But not the development of embedded devices.

So what about the tinkerers? (what's the collective noun of 'tinkers'? A gadget of tinkers?)
What is it that they do?? It isn't Business I.T. - but is often done on the same hardware/equipment.
These are the people that would've once been 'ham' or amateur radio operators.
My local Linux Users Group is where I find them now - or LCA - Linux.Conf.AU - if I was prepared to pay ;-)

Business Systems, are constructed from all these components.
In Business I.T., a chief concern should be for The Data.
If you upgrade MS-Word or MS-Excel and can't access old documents, is that important to someone??

If your business includes such documents in a critical business process/system, it's now bought a whole whirl of cost and trouble "migrating" them.

New Dimensions
  • Appliances vs General Purpose Computing Devices (fixed/single or general purpose devices?)
  • Desktop/client Applications vs Server Applications (Where does your App. run?)
  • Infrastructure vs Server & end-user devices (Invisible or visible to the user?)
  • Services vs Applications (indirect or direct user app.)

Implications for Design, Operations and Performance Analysis


Design
  • Appliances: are "fungible". They perform relatively fixed, simple functions and conform to standards. Managers can compare specs and prices meaningfully. They also depreciate in a more understandable way. Today's high-end NAS appliance will still be useful in a role in 5 years. Ten year-old routers may still work and handle useful traffic.


Operations
A manin objective of I.T. Operations and Admin should be to invent themselves out of a job. So they can get on with more important and valuable work. Automation is the name of the game.

We don't have "Tape Operators" any more. Those jobs have been automated away.

We need some metrics to quantify this:
  • Number of operators/admins per 100 users
  • Number of servers per operator/admin
  • Number of Gigabytes per operator/admin
  • number of databases * Gigabytes per admin


And my favourite, the "R-Mark":
How big an office, desktops, servers and services, can a informed non-professional set up with a credit card and the local computer 'superstore'.

The "R-mark-II" is:
How many networked offices


Performance Analysis

100% Availability of networked services is becoming more easily and cheaply achievable.
It's a target that can be generally aspired to.
And hardware is becoming 'so cheap', that for most sites, there should always be enough capacity.
And for those that can afford it, we should be seeing "Datacentre in a Box" solutions soon.
[i.e. a shipping container fully loaded with servers and disk. Maintain once a year, if at all.]

In this world, do we have a greater or lesser need for performance analysis and capactity planning?

Other things
Don't have time to expand on all these just now...

appliances, rich/thin clients, infrastructure, services

No comments: