In a post last week, I wrote about the need for small and medium businesses to plan their IT strategy in such a way to position their organizations for expansion. To help readers understand of the issue, I highlighted specific areas where an SMB can exercise better discernment.
Today, I would like to examine another couple of areas that I did not manage to cover at that time.
The pressure is often great to keep costs low in an SMB. And more than any other area, the temptation to scrimp is often greatest when it comes to network equipment -- the supporting role it plays means that it is also the least visible item as far as management is concerned. Least you be tempted, I have earlier made my point as to why consumer-grade networking hardware often does not cut it in a corporate or mission-critical environment.
For networks larger than 24 nodes, I would strongly advocate that the SMB consider purchasing a network chassis rather than a vanilla switch. As an example, I once opted for the HP ProCurve 4202vl 72 switch for a network with about 65 nodes. This gives me room for expanding the network by another 48 ports by purchasing two 24-Port 10/100-TX modules, add in a combination of modules with 1GbE ports or 10GbE ports, or even snap in the capability to link to another remote switch via fiber optics.
In addition, its innate capabilities to perform VLAN mean that I can create virtual LANs for the purpose of testing or isolating selected equipment - without even having to unplug any equipment or purchase another switch. While your preferred networking vendor might vary, the important thing is to cast your eyes a little further and build your network with more than immediate needs in mind.
Servers used to represent the most challenging, and expensive, piece of equipment to buy. Compounding this was the fact that a new server can take weeks to arrive, which can be problematic in many cases. The advent of virtualization not only mitigated this problem, but simplified matters tremendously.
I once had a Managing Director insist on getting the "best" server possible to replace a machine that suffered a catastrophic hard disk crash. It was a pity because the budget for this single -- and way over-powered -- machine would have gone a long way towards the virtualization plan I had drawn up to address disaster recovery and business continuity concerns.
My point here is that the assigning of servers should no longer be role-based. Of course, security and legal considerations might mean that the physical servers hosting the software or virtual machines for the accounts department cannot be used run the company's Web server. These considerations aside, it would be a grave mistake to apply limited, traditional thinking against the technologies that are currently available.