Cost Justification of (technology) Systems upgrades / expansions

Introduction:

In my many years in Motorola and in the consulting industry since I left Motorola, I have had numerous chances to both define a short-fall in existing technology, and prepare a plan and justifcation for repairing it.  I have also created a number of new plans for the  expansion of computing and network operations.

For whatever reasons, I rarely fail to get the dollars requested, in order to accomplish expansion and upgrade plans or new-facilitization plans.

I'll share with you, here, through both theory as well as some examples some of guidelines to follow to make a successful purchase justfication to management

Background:

Many times, especially if you are a newly hired systems, networks and security administration manager you've been hired into a less than adequately funded or supported networked computing environment. This is not a reflection on you, but on any company's ability to not focus where it should, and consequently to have less-than-adequate computing facilities.  It is also a result of people and departments and management that does NOT do an ' ANAR ' every year in preparation for budget-time!

There are a lot of reasons - none particularly valid - for having a less than adequate computing environment. My role here is, though, to help you be successful in creating an expenditure proposal to upgrade your environment to where it becomes productive, efficient and economical to support your user community so that they can support their ultimate customer-base and we can continue to grow and prosper.

First-things-first:

One of the key items in developing an expenditure proposal is to gather data to validate the existing state of the computing and networking environment.   But first - lets make sure, through a "DOT" (Definition of Terms) that we all agree on what some words & phrases mean:

Computing & Networking environments consist of at minimum these parts:

• WAN (Wide Area Networking (country or city to country or city)) Connection
• LAN -WAN Interface: firewall, router(s), XLANs (connection to non-Motorola entities)
• LAN-infrastructure: (internal) router(s), switches, hubs, and cable plant
• SERVERfarms: applications(GNU, Frame, Clearcase, etc.) & network services (DNS, email, time, file, directory, etc. etc.)
• Desktop computing: computers that utilize a desktop CPU & a desktop hard disk; these machines may do software development work, office-automation work (spreadsheet, word-processing, database, and the like and/or may be utilized solely as "thin clients" (see below)
• Thin-client: X11 terminals, both dedicated as well as desktop PC's running X11 that enables a PC to access X-11 applications.
• Portable computing: laptop computers that some employees will use only when travelling, yet some employees use as their only computing platform.
• LAN-support-infrastructure: this is my name (not very good) for the "catchall" area that includes items such as backup tape drives, CD-writers(for archiving), CD-ROM Changers (some call them "juke boxes") for on-line storage (typically) of documentation, network & computer monitoring & management tools (hardware & software) and other items that are typically used to support the entire environment, but do not fall in any other category.

Let's assume - for the sake of brevity, that you agree with and approve of my above deliniations. Then we can see that in "most" cases it will be fairly painless to convince management that if you're going to hire 100 new people next year, (you already have 100 users), that you're going to probably have to buy / spend on the following things:

• WAN link expansion
• switch or switches and/or new hub or hubs, perhaps new cabling
• LAN punch-down-blocks and perhaps a new rack to put them in
• perhaps a new or expanded 'data closet'
  
• server hardware and server software to say nothing of user-application(s) software
• about 100 new desktop, thin-client or portable computers
• tape drives or other items to expand your back-up system to accomodate the data these new employees will generate

In most cases I think we can all figure that there's little "discussion" on justification for such purchases. On the other hand, if you don't see this and/or don't agree, write-call, we'll discuss.

However, let's now paint a different scenario. You're a new admin, you've just been hired into Company A at site "Z", and you feel, somewhat without existing data, (because you've just been hired) that this network just isn't cutting it and there is significant room for improvement - perhaps with fairly small expenditures.

Lets also assume that you don't have the tools and techniques to evaluate the environment's parameters, but you need some help.

Evaluate the problem-set:

1.) WHERE is the performance problem, which of these is causing problems:

• WAN
• WAN-LAN interface
• LAN infrastructure
• Servers
• dekstops
• portables
• thin-clients (almost never a problem)

In almost all cases, a severe performance degradation is a function of more than one of these items.

So, research & ISOLATE the problem component or components, FIRST.

2.) WHAT is the performance problem?

• related to problematic hardware, software, OR a combination of the two
• related to configuration of hardware, software OR a combination of the two
• related to random (as opposed to planned) network growth?

3.) HOW can we fix it? - (in increasing order of costs in dollars & time)

• reconfigure existing hardware / software? (perhaps expense dollars)
• upgrade existing hardware / software? (capital dollars)
• replace exisiting hardware / software?
• additional hardware / software?
• additional staff?

4.) HOW SUCCESSFUL (or long-term) will the "fix" last? - is this just a tactical solution (less than 12 months) or will we get more value out of this? - This is the ultimate decision maker here. How much return vs how much cost?

5.) ASSUMING you are able to identify, isolate, and document either the problem and/or the problem and it's solution, you should now have a good basis for how much it will cost to fix or improve the situation in the future.

Now - that we have figured out where the problem is, and we have a potential solution, we must decide how to justify the expenditure.

Data is your friend. With data about the state of the network today (available with some time and the right tools) as well as the state of the network a year ago, you then project the performance issues into the future. Comparing the past, existing and future data projections against the number of users / employees, and you will be able to show a difference between where you are today (good / bad or neither) and where you want to be in the future.

Tie data to performance:

Now - the key here is to develop the data - indicating a performance trend, and tie that performance trend to the dollars you want to spend.

Computers are a productivity enhancement tool:

Also remember, though, that a computer networing environment is here for the employee's use because it is supposed to be a "productivity enchancement tool". Remember this very clearly. This is the only reason the company buys & puts computers in front of us. They are not there to send e-mail to our wives & girlfriends, get the news or sports scores, or look at our stock portfolio. They are there to make us more productive employees so that we can generate profit for the company at a higher rate than our competitiors. Remember this key item, and that management looks at computer expenditures form this standpoint, and your justification will be easier.

In order to "sell" your proposals for spending dollars, upgrading equipment, adding new equipment, and hiring new staff, you have to remember that our management is driven by little else in this world except customer hugs & dollars. We can't do a lot at our level about customer hugs, but we can, if we're careful, AND, more importantly if we PLAN AHEAD, collect data & use that data to support our case.

Remember, if we do this Annual Network Architecture Review, on an "annual" basis that means we should be able, this year, to use last year's data, plus 'current situation' data in order to project either a short-fall or a need to do whatever our users, networks & customers need us to do, into the future.

However, management doesn't have that warm & fuzzy feeling that you do for that new Enterprise 5000 warming your feet on a cold winter's day, all he/she/it sees is dollars that are not in the profit column, SO: you must put this purchase in terms he/she/it can understand:

If you want new hardware/software - you have to couch the purchase in terms of increases of productivity, because that typically translates into more customer hugs and more sales, which makes management VERY HAPPY...

Cost Justification, an example:

You've got an old Sparc20 / 21" / 2 x 2 gig internal / 96Mbytes that's been doing DNS & SMTP-POP-IMAP duty for the past three years, but you're regularly filling-up /var/spool/mail, and you've had 2 x hard-disk failures in the last 8 months, so you're "nervous" about this machine's future. You've grown from 50 to 200 users in this time, and you think the machine may need to be either upgraded or replaced.

So - first thing: find out how bad the situation really is:

• Run CPU-load (such as Xload) on the machine for a couple of weeks - collect data.
• Recover from archives the CPU-load data you ran a year ago (you DID run that a year ago, right?)
• Do the same thing for memory-useage & hard-disk traffic
• Review your mail-log-files for excessive "delay" parameters on your incomming and/or outgoing mail
  • (Is the machine REALLY having trouble, or does someone just perceive this?)
  • collectively - what does this data tell you?

IF you feel there's a real problem here (that, for instance, can't be solved with a hard-disk / memory upgrade (and a good cleaning in most cases (when was the last time ANY of you took all your machines apart and either blew-out all the dust & dirt with an air-compreesor, or sucked it all up with a vacuum cleaner?)) - Got a machine crashing, periodically, for no reason - it's probably nothingmore than dirty? - try it - you'll like it!