Just Say No to White Boxes in the SMB

Posted on
Like what you see? Share it.Share on Google+Share on LinkedInShare on FacebookShare on RedditTweet about this on TwitterEmail this to someone

One of the most challenging things about small business consulting is trying to get your clients to be open minded about using the same technologies that larger enterprises use. You always hear things such as, “We aren’t the CIA” when it comes to security, or “My brother’s friend said he can build us a machine for cheaper”

Both statements are true. But your brother’s friend might not be able to make himself available when something goes wrong with that machine.

Now, I am not saying that white boxes are not a good option. For many companies they are a cheaper alternative to get workstations, and as long as there are no major problems with them, a small business can save some decent money going with these for workstations. Major manufacturers such as Dell and HP have recognized this market and now price machines to compete with these local retailers. This is probably one of the best things that could happen, as every small business really should be purchasing machines with a warranty and a built-in support system. Many larger companies can afford to have “spare” machines around in the case of a failure, but for a small business, a crashed or downed workstation can mean lost productivity of employees, and if it’s your accounting or payroll person’s machine, it could mean the whole business slowing down.

The real problem I have with white boxes is with servers. I know for a fact that anyone can go to Newegg.com and build a workstation that will compete with and be priced lower than any HP Proliant, IBM xSeries or Dell SC series Poweredge. Typically, my argument against doing this is the lack of support. HP, IBM and Dell are all major Microsoft partners. Everything in these servers has been tested and is on the Microsoft Windows Hardware Compatibility List (HCL). Additionally, when people build their own servers, there is no real warranty. There might be warranties on individual parts and components, but if the machine decides to just start Blue Screening one day, you have a tremendous task on your hands trying to find out why. Is it bad RAM, maybe an incompatible motherboard, perhaps the wrong jumper settings on the IDE hard drives?

This is where the term total cost of ownership (TCO) comes in. I used to have a client that would consistently pay about $1,000 for his white-box servers. One day, one crashed with a BSOD, followed by a corrupt registry error. I spent approximately six hours trying to repair this, including purchasing new hardware to replace suspect hardware. I have to admit, he got the server for cheap. When I cracked it open, I was shocked to encounter a Gigabyte motherboard, AMD Duron Processor, RAM, NIC, Video card and a single IDE drive. Maybe the $1,000 was not even that great a deal after inspecting the components involved, but it’s still cheap for any machine considered to be a server. The total cost for me to attempt recovery of their server, and to put a new solution in place will end up being considerably more than that. Granted, looking at server purchases this way, a skeptical small business owner can still say to me, “Well, we got two good years out of it.” Luckily the server in question in the above example was not one that many people relied upon.

The trick to getting an SMB customer to realize the value of purchasing a better server solution is not an easy one at all. The key in my opinion is the TCO model. You pay up front rather than paying more later in an emergency fashion.

Let’s talk worst-case scenario for a moment. Picture the same server described above, used as a client’s single server. They have their Quickbooks data, all corporate data on the server, as well as their custom Access based database system that holds all their order processing information. Additionally, the server is their single and sole domain controller for their domain.

That same server dies in a similar fashion, BSOD, most likely tied to hardware failure (most likely because the hardware used to build the machine was not hardware intended for server usage)

As a technology consultant, you walk into this situation. A new client calls. They are down, this is what you find. You are going to help them out as best you can, but first, you need to get their data to a safe place. Let’s assume the single IDE drive has not failed, and you can copy the data to another location. You are still going to spend a good amount of time copying that and recreating shares. You will then need to recreate their domain and give them access to that data. Let’s assume this will take at least 10 hours at $100 per hour (for a single technician to complete all the work) that’s $1000 right there, and we haven’t even talked about replacing the server yet!

Therein lies the trick. You need to make your clients understand the risks they face and the costs behind them. Sometimes it is hard to do this without sounding like you are trying to scare them into a purchase, so you need to be careful with what you say.

The biggest reason in my book to go with a major manufacturer for a server is the 24×7 support. And with many of them, they have a four-hour response time for replacement parts. This means that if a drive crashes, or a motherboard fries, you have a technician onsite in less than four hours with a new one. Add to that all hardware is compatible with your operating system and the additional redundant features such as redundant power and cooling sub-systems.

Also, in my example above, you might have noticed that this particular client was using a single IDE drive. No RAID (Redundant Array of Independent Disks) solution means a single hard drive failure could mean the loss of all of their data, as well as the operating system’s boot disk, rendering the machine pretty much useless. Also, when many companies build these white boxes, they are in the business of making money. That means the cheapest components they can find for the lowest price. They essentially use the same motherboard, RAM and drives that they would on a machine intended to be a workstation. The overall quality of these components may be good, but they are not intended to be running 24 hours a day, seven days a week and serving 50 users simultaneously.

Over the past 10 years, the MBTF (Mean Time Between Failures) on hard drives and retail PC components has most likely gone down, as the turnover time on new computers has also decreased. The amortization period for computers used to be five years, but around 2000 we saw many accounting firms start amortizing at three years, as they were finding they were upgrading the technology more often. This change has obviously led to retail components being engineered to last a shorter time period as well by using lesser quality materials, because as a whole, the industry wants users to be upgrading to the newer equipment.

This is the exact reason why a machine built with retail parts should not be a business solution. When a company such as HP plans on building a new server, it plans for a few things. First off, all components work together, talk to each other, have built in logging functionality to a single location (HP’s servers have an Integrated Management Log that list any hardware problems in an easy to read format) and more importantly, they’re built to scale. You can add more processing power, more RAM, multiple SCSI cards for data storage devices or backup devices, multiple bays for DVD writers, CD writers or tape drives and much more. They also include some nice-to-have features such as physical security in the form of locking cases and enclosures that can even report the last time the cases were opened in the integrated logs.

Although these reasons might not be everything you need to convince your clients that a new server purchase is the best bet for them, it should at least be enough to get them thinking about what they have spent in the past. Don’t be afraid to ask them what they have spen

Like what you see? Share it.Share on Google+Share on LinkedInShare on FacebookShare on RedditTweet about this on TwitterEmail this to someone


Posted in Archive|