Data Center Configuration Best Practices
Following Configuration Recommendations
Most OEMs provide documentation on best practices for installing and configuring their product offerings. But are they truly necessary? Will they actually keep your devices happy, optimizing performance and preventing unnecessary outages?
The quick answer - yes. It's a bit like reading the directions before putting together that complicated RTA furniture. Sure, you might be a superstar, and the piece of furniture is upright (mostly), but we all know you had a couple of mysterious screws leftover 😉
While many best practice recommendations seem fairly straightforward and obvious, others can seem a little pointless or even a pain to implement (especially if they require a system shutdown). But we've seen countless environments with, um, let's call them, unique configurations. And we've witnessed (and fixed) many issues that could have been prevented.
While our advice is to follow ALL of the OEM best practices for installations and configurations, we've gathered a few top offenders—recommendations that are often ignored and cause problems for people.
- Dangers of Overheating
- What Else Can Go Wrong
- Airflow Quick Tips
- Common Cabling Mistakes & The Consequences
- Cabling Configuration Tips
- Redundancy Configuration Blunders
- Redundancy Tips
Airflow in the Data Center
Thermal & airflow issues are among the top causes for failures plaguing data centers. And even though most of us working in IT know that proper airflow is important, following best practices is still a struggle. Let's take a look at what can go wrong and a few general guidelines to keep in mind.
Keeping It Cool: The Dangers Of Overheating
Servers, storage, and networking devices all need to keep their cool. When things get hot, all kinds of fun things can happen.
If you're lucky, you might find that performance tanks (many devices automatically reduce speeds to generate less heat) or you may experience unplanned shutdowns (again, an automatic response when they reach a certain temperature, preventing more permanent damage). But you might also see more dramatic consequences, like damage to hardware components (think melting, warping, sparking, or even exploding of CPUs, hard drives, switches, etc.), data loss, and even shortened lifespans (how much were those brand new servers??)
What Else Can Go Wrong
While overheating is the most obvious airflow issue, other problems can happen from neglecting airflow best practices.
Dust!
Poor airflow can lead to dust buildup, which can cause fires and clog (and even break) fans.
Humidity, condensation, & moisture, oh my!
It doesn't take SME level expertise to understand that moisture around data center equipment isn't ideal. Poor airflow can cause condensation in cold spots, corrosion, and even sparks.
Shake rattle & roll!
Vibrations from overworked fans can cause all kinds of problems, including loose connections and even data loss (hard drives don't love being shaken up constantly, but who would??)
Lights out!
PSU failures from dust and triggered breakers can cause power outages.
MGS Airflow Quick Check Tips
It's always a good idea to follow the specific guidelines for each device, and there may be specific recommendations for servers, storage, or networking devices. OEMs may also have brand-specific designs, labeling protocols, and parts.
Here are a few general best practice concepts to keep in mind.
Make sure cold air is coming in through the front and hot air out the back.
Equipment needs to be able to "breathe". They "inhale" cold air at the front of the device and "exhale" hot air through the back. Hot air sneaking out the front can cause problems.
Use blanks in empty spaces as per the OEM recommendations.
Gaps can create airflow leakage.
Tidy up the cables on the sides.
Keep the front and back clear of cables so they don't block airflow.
Match the arrows.
Blue arrow = normal (front>back); red arrow= reverse (back>front)
Check fans.
Make sure all fans are spinning, and there are no red lights.
Cabling Configuration
Another top area we see configuration recommendations ignored is cabling. It's pretty obvious that there is much more to cabling than simply plug and play, but many people don't take it seriously enough.
Not taking the time to set yourself up for success may mean that future you will be dealing with a giant pain in the rear.
Common Cabling Mistakes & The Consequences
Not using the recommended cable type
This could include specific cable types, connectors, & lengths, and more.
Consequences:
Power issues (voltage drops, overheating, etc.), slower speeds (i.e. high speed networks - if the cable isn't rated for it, you're not going to get the expected speeds), problems with the signal (think dropped connections, reduced bandwidth, etc.), and even fire hazards (arcing, electrical shock risks, sparking, etc.)
Substandard Cable Routing & Management
This could include sharp bends, crowding cables, missing retention clips, and loose connections.
Consequences:
- Excessively bent cables = damaged cables (the inner wires can get jacked up)
- Crushed or congested cables = heat buildup, failures, signal interference
- No click = loose connections which eventually = no connection
Lazy (or worse, no) Labeling
Not clearly marking cables at both ends may take a little time on the front end, but if you don't...
Consequence:
A super fun guessing game when it comes time to do any cabling troubleshooting, adding hours onto your project. It goes something like this: "Let me just chase this cable down...son of a...sea biscuit! Where the...heck??" or possibly something even more colorful.
Not Planning For The Future
Ignoring scalability and future-proofing concerns will come back to bite you.
Consequence:
It's a blast to redo what you just did, right? And cost effective, no? Things move so fast in the technology world that if you aren't planning for the future, you are already behind.
Cabling Configuration Tips
Label cables at both ends.
Extra credit for color coding (by function or device).
Map it.
Use a spreadsheet, map, or both. Future you will be so thankful.
Leave breathing room.
Having extra space & capacity (ports, trays, etc.) ensures easier future expansion and cable health.
Keep it tidy.
Dealing with a rat's nest of cables isn't just a pain when you have to replace or troubleshoot the them. There's a higher chance you'll have restricted airflow, loose connections, bends, and maybe even an actual rat's nest (shudder).
Ignoring redundancy is a bit like high-stakes roulette — spin the wheel and see if your data is safe. Hmm. Maybe not the best idea.
We have seen problems magnified and become critical because of insufficient redundancy, whereas in other environments they barely caused a ruffle because of additional devices, proper backups, and other redundancy strategies
We obviously have strong feelings about the importance of redundancy — we made a whole page dedicated to redundancy strategies!
In some cases, guidelines can cross categories. For instance, in some 3PAR devices, failing to follow certain cabling guidelines can render any other redundancy plans you may have put in place irrelevant. As always, it's essential to follow the OEM-specific protocols that apply to your devices. However, there are still some common steps people ignore.
Redundancy Configuration Blunders
Not testing (systems, backups, power supplies, etc.)
This seems like such a simple step, but it's one of the most common mistakes. We've seen this countless times. In some ways, a false sense of security can be worse than knowing that you are taking risks.
Inconsistent redundancy execution
Focusing on only one part of your system and ignoring others can leave vulnerabilities. The devices all work together and are part of the same environment. Every redundancy plan is only as strong as its weakest link. For instance, you might have an amazing strategy for your networking gear with multiple switches and cabling that slaps, but if you've neglected to make any redundancy plan for your storage devices, failures will inevitably catch up to you at some point.
Sharing resources across redundant devices
Any failure in shared components can break your redundancy. An example of this would be two redundant servers in the same rack sharing a PDU and linked to the same switch.
MGS Redundancy Tips
Spread the power. 2x2x2
2x2x2 = Two power supply cables plugged into two different power strips plugged into two different power sources. Failures will be less likely to cause problems.
More is more.
Distribute resources, divide the workload, and use multiple devices. Don't put everything on one machine. For instance, if you have 24 drives, put 12 in one enclosure and 12 in another.
Test backups and spare devices regularly.
It's not truly redundant unless you know for certain that it works!
Get your disk redundancy in order.
Use at least RAID 2, but RAID 5 or RAID 6 is better (learn about RAID groups)
Replace failed parts as soon as possible.
Don't wait to replace that failed hard drive, power supply, fan, or whatever just because you have a backup. You are just asking for a disaster!
Don't get complacent.
Systems can change, hardware can fail and staying on top of both your strategy, and the hardware is a constant process. Don't think of it as one and done.
Let M Global Help
We love helping our customers get their configuration wrinkles ironed out. It's a win-win for both of us. Optimized configurations mean longer lifespans, better performance, less downtime for our customers and fewer headaches for us during service calls.
We want you to consider us a trusted resource, advisor and an extension of your team. We're all about helping to solve your hardware problems.
Call us today at 888-675-1895, fill out the form, or start a chat with us to get the conversation going.
Contact Us Today
Suggested Content
M Global Services Offerings
Third Party maintenance support is what we do. We live and breathe hardware support. But we can offer a few other additional services if you need them.
More to Explore