Servers

How often should enterprise admins replace Mac servers?

Erik Eckel recommends a Mac server replacement plan that takes into account the cost of retaining machines that are past their prime.

The question of how often enterprise administrators should replace Mac servers is easy to answer. Justification is the tricky part; finance and accounting types might not like the answer.

A few weeks ago I presented four questions to ask to determine how often Mac workstations should be replaced. Mac servers possess vastly different lifecycle requirements, however.

Mac server v. workstation replacement cycles

There are at least three reasons Mac servers require replacement at different intervals than do desktop workstations:

  1. Mac servers typically fulfill more demanding video production and/or audio editing tasks than do Windows servers. Over time, new Mac server models boast improved architectures that introduce new performance capabilities.
  2. Mac servers, by nature of the video, audio and graphics work they frequently empower, typically host vast amounts of data. As a server progresses through its service life, it's not uncommon for disk space and storage issues to arise.
  3. Like Windows and Linux servers, Mac servers must deliver near uninterrupted uptime. The servers cannot crash or business operations and production are interrupted.

The answer

So just how often should enterprise administrators replace Mac servers? I say every five years. Organizations should build their budgets with server replacement planned twice a decade.

It makes no sense to try stretching life cycles longer, especially as accounting statistics suggest it costs more to maintain systems the more they age. Hard drives fail. Power suppliers fail. Motherboards fail.

The more hard disks, the higher the likelihood of failure. Due to the demanding tasks Mac servers empower, it's likely the servers utilize multiple hard disks. Even the Mac Mini Snow Leopard Server ships with two 500GB SATA drives, while Xservers can be configured to include up to 16 SATA or SAS disks using optional Promise VTrak E-Class RAID controllers.

Waiting's cost

Wait too long, and enterprise administrators will find themselves having to waste IT budget dollars for repairs to systems that should be being rotated out of the organization. That's not a solution; that's a temporary fix. And it's expensive. By cycling Mac servers every five years, organizations can ensure they do all they can to eliminate unplanned downtime, while also benefiting from faster architectures, greater amounts of storage and improved reliability.

The true cost of downtime, especially unplanned outages, is simply staggering. The formula is surprisingly non-complex. Just divide annual earnings by the number of workdays in a year and then subtract the amount of downtime from the available workdays.

For example, say an organization with just $10 million in annual sales suffers a server issue that requires just a single business day to repair, which is unlikely, as diagnosis and obtaining replacement parts typically requires at least one business day, with a second day being required to actually perform and test the repair. During the day the organization is down, due to the server being unavailable employees likely can't work, complete projects, process orders, send invoices, respond to email, etc. That means the organization just lost $40,000.

Skeptical? Check my math at $10,000,000 annual sales divided by 260 weekdays a year minus 10 federal holidays (250 workdays total) equals $40,000 a day lost due to downtime. Adding insult to injury, now the organization must invest precious IT dollars band-aiding a server that's just going to experience more difficulty as it continues aging.

Those figures should send a sinister shiver up the spine of the accounting and finance officers. Enterprise administrators can help ensure organizations avoid such catastrophes by ensuring IT budgets call for replacing server chassis every five years.

About

Erik Eckel owns and operates two technology companies. As a managing partner with Louisville Geek, he works daily as an IT consultant to assist small businesses in overcoming technology challenges and maximizing IT investments. He is also president o...

8 comments
don_sanders
don_sanders

Replace it once with something else :)

Jaqui
Jaqui

lost income of 40,00, PLUS the wages for he staff to sit around doing nothing with no income make it higher. it's not just the income, the expenses continue making the losses greater than just the income.

wizard57m-cnet
wizard57m-cnet

Since in the USA, the IRS has set depreciation tables for electronic equipment, including computers, at 5 years, it's a "no-brainer" to recommend replacing ANY server system at 5 year intervals.

kenrblan1901
kenrblan1901

A Mac server is generally taking on the same types of responsibilities as ones running Linux, Windows or Unix. The hardware is very similar if not identical in its composition. SATA and SCSI disks have generally expected duty cycles. The same applies to the other components. Like other servers, replacement parts become more difficult to acquire beyond that 5 year window. In addition to the concerns about the hardware durability, you have to consider the productivity gains and possible energy savings associated with newer servers. When I worked for a university, we only had a couple of Mac servers. The reason we didn't buy more was simple. They simply didn't offer redundant power supplies and weren't fully supported to run VMware ESX. Everytime I saw our Apple sales or technical reps, I would ask when they were going to be serious about the enterprise market. I loved their workstation, desktop, and laptop systems and wanted that same reliability in the server room. I also wanted the option of legally running OSX Server in a VM. Of course, we also beat the reps up over not offering a tablet system, but Apple finally came around on that one.

tbmay
tbmay

...the same thing about all their servers. I care not what it runs, if it has moving parts, somethings going to go belly-up eventually, as you indicated in your article.

nagendra_pratap_singh
nagendra_pratap_singh

Replace it with a Dell. They have redundant power supplies, cost less and take up less space.

C.Daly
C.Daly

Your math would imply that if my credit card bill is just about to be printed and the Visa server goes down, I won't get a bill and I won't have to pay this month. In fact, I think your math says I won the lottery and won't ever have to pay this month's balance? The income would be lost... That server activity would just be resumed another time. There might be a dip in the cash flow due to the delay mailing bills, followed by a compensating rise shortly thereafter. Part of the cost we pay for server failures is inconvenience. Truly time-critical computing like processing the credit charge for my breakfast is outside the scope of this discussion because businesses buy redundancy and infrastructure to handle failures with no down-time.

Editor's Picks