Intel [is] Inside White Box Servers, too!
Before we dive into our list of 7 server hardware alternatives, let’s start with critical thought.
Here’s what we all know:
Google and Amazon build their clouds on customized server platforms with motherboards, disk drives, and backup batteries in open enclosures instead of fancy name brand cases with cool face-plates and someone’s logo.
Low-Cost White Box Servers…
To cloud providers like Google and Amazon; compute, space, bandwidth, efficiency and lower infrastructure costs are key to their success.
Let that sink in…
I’m not saying we should all build our own servers but do we need most of the features we’re buying?
Sure if you want whistles and bells [which you may never have time to correctly set up because your Ops team is too busy working on day-to-day projects] then Buy UCS or another top brand.
On the other hand, if all you need is compute for VMware virtualization that can be swapped out when something dies, aren’t white box servers good enough?
DIY vs Name Brand vs White Box Servers vs Refurbs
Before we’re done with this Introduction to Virtual Server Guide, I’m going to provide a list of brand-name server manufacturers, hardware parts depots for open compute projects, and white box server alternatives you can use to source your next project.
Chips are chips…
Intel makes their money on chips, and they pretty much sell the same chips to everyone.
Yes, I know not all hardware is made the same…but that’s why we build clusters and clouds that can do vMotion, Live Motion, or XenMotion.
Ah, the benefits of modern technology!
vSphere and other alternative virtualization platforms are perfect for low-cost hardware so before you let vendors talk you into the best of brand hardware and support agreements, consider the alternative.
Unless you’re NASA and lives depend on the chips in your servers, you should consider lower cost server hardware alternatives.
Quick Video About The Open Compute Project at Facebook
It doesn’t need to be pretty! It doesn’t need to be painted! It doesn’t need to have plastic bezels that guide airflow around! – Johathan Heiliger
And here are a couple more reasons to consider open compute…
If your hardware architects have done their job and have built-in room for failure (n + ?).
And your software architect has designed your app to work across multiple servers in an App pool (preferably each on a different host), then you can take an outage on a VM, disk, motherboard, NIC, switch, load balancer or host, and your app will still be online.
Now granted, depending on the sophistication of your App some users may get bounced out, but that’s a fixable problem. And these problems can be solved with code instead of throwing more costly hardware at them.
Now:
Let’s get to what brought you here. Server hardware alternatives for building clouds.
7 Server Hardware Alternatives (updated to 7.x options)
Best of Brand Server Options for VMware Virtualization:
1. IBM Servers (Now Lenovo) – Â Long before HP and Dell, IBM was building server and desktop hardware. And over the years I’ve had my share of opportunities for supporting iSeries hardware running AIX, and xSeries running Windows, Linux, and VMware. No complaints, however, you will pay for the brand and higher quality. Click Here >> To View IBM Server Hardware Options
2. HP Servers – Once upon a time there was Compaq, And they were purchased by HP. And now we have HP Proliant servers. HP is another hardware platform I’ve spent a lot of time on. HP rack servers and blades are great for running raw metal servers with Windows and Linux, or for building high-density virtualization server infrastructures. I’ve succeeded with c7000 chassis and BL460 g6/g7 blades and DL380 G6/7/8 and DL580 G6/7 running VMware vSphere. The only real problem I recall was managing firmware upgrades and the frustration caused by generation changes. Â Click Here >> To View HP Server Hardware Options
3. Dell Servers – I’ll be honest. As an alternative to going all the way white box, I would take Dell PowerEdge server hardware all day long. Why? Because Dell makes great general-purpose servers that can do it all. And pound for pound a Dell server is as good as IBM and HP. They also make a great 1U pizza box server that is perfect for running ESXi. Another value-add is Dell has a sweet spot for lower TCO when building a vSphere cloud over other top brands. Click Here >> To View Dell Server Hardware
4. Cisco UCS – If you have a Cisco network, I can bet your network team is running a rogue Cisco UCS POC right under your nose. How can I say such a thing? Because I’ve seen it happen in 3 places, I’ve worked. Cisco is notorious for giving away UCS chassis and blades to seed businesses with their server hardware. The problem I see here is long-term costs will go up, and you will find yourself refereeing network and server teams that both feel they own the UCS platform. My Beef here is I’m opposed to vendor LOCK-IN, and Cisco is at the top of the list. Click Here >> To View Cisco UCS Servers hardware
If you’re hosting your applications in the cloud, chances are you’re running on white box hardware or a Franken-Server similar to what’s in the picture above!
White Box Server Vendors (ODM Server Vendors) – Refurbished Server Hardware & Parts:
5. Super Micro Servers – Super Micro has been around for a while now and honestly, I’ve never had the opportunity of supporting one of their pieces of hardware. It could be due to the lock other big server brands have, or the stigma of saying you are running on Super Micro? I guess it would be like saying you drive a Hyundai instead of Honda. For the record, Super Micro has a large selection of server hardware to choose from with a similar configuration to Dell and HP. And they also have a blade and chassis for high-density computing. Click Here >> To View Super Micro Server Hardware
6. ABMX Servers – Now we are talking true white box servers. On the ABMX website, you can choose from a variety of 1u, 2u, 3u, 4u, and blade server hardware configurations. Pick any drive, controller, CPU, PS, or memory combination from a drop-down. Just like a candy store at the mall, ABMX has everything a server hardware guru could ever dream of. ABMX was founded in 1994 and is located in Orange County California. Click Here To View ABMX Server Hardware
7. Servers Direct – Another truly white box server outlet that has it all just like ABMX, except they offer a variety of DIY barebones and custom systems. And Servers Direct resells SuperMicro Servers and other white box brands. With a fantastic white box server inventory to choose from, Servers Direct is also located in Sunny California, and has been doing business for 10 years. Click Here >> To View Build Your Own Server Hardware
7.1. Build Your Own – Open Compute Designs are becoming more popular, and the Open Compute Project is the place to visit for ideas on open server, storage, network, and data center designs. Click Here >> To Visit the Open Compute Project.
7.2 Iron Systems  – I recently started purchasing all my whitebox servers from Kawal at Iron Systems. They build a robust white box server made from Super Micro parts. I have been pleased with their servers for building ESXi and Xen clusters. They offer low-cost servers, as well as white box network and storage products. Click here >> To visit Iron Systems and ask for Kawal or email him: [email protected]
Here is an in-depth guide if you’d like to learn more about whitebox server hardware advantages and disadvantages.
7.4 Hard Drives Direct – Low-cost brand name server parts. Drives, Memory, CPUs, Power Supplies, Controllers, and more. Click Here >> To Visit Hard Drives Direct
7.5 Server Monkey – Server Monkey is another option for purchasing refurbished server hardware and parts that have been used and then restored to its original condition. Click Here >> To Visit Server Monkey
More About The Facebook Open Compute Server
We design the server to be very friendly for the technician so it’s easily swap components in and out. – Amir Michael
Wrap Up
Well, there you have them. Seven brands and vendors offer too many server options to compare or review. And each has hardware solutions that are perfect for your budget or ego. We covered server hardware from the best of brands manufacturers, white boxes that are low cost and cheap, and open compute servers.
Maybe your budget isn’t what it used to be and it’s time to consider switching from costly hardware to low cost servers that offer the same Intel computing power as the big brands?
Let me leave you with a final thought to process before you buy…
Given most chips and components are made by the same people in the same factories; the real differences you are paying for are brand name, features, and support.
So let’s look further into support – most major brands are moving their support off-shore to the same outsourcing companies. Hmm…
Now we are down to features and brand name.
Do you want to pay 20 – 40% for features you won’t use? No.
And now we’re down to brand name:
From what I’ve found, Google, Facebook, and Amazon don’t seem to mind using open compute or white box servers. Thanks for your interest.
Read Related Topics:
VMinstall.com is a participant in the Amazon Services LLC Associates Program, an affiliate advertising program designed to provide a means for sites to earn advertising fees by advertising and linking to Amazon.com, Amazon.co.uk, Amazon.ca, and other Amazon stores worldwide. *Best Sellers last updated on 2024-10-07 at 01:08.
Thank you, finally someone who gets it.
When was this article published? I’m sure the preferred order of these options has changed over the last few years.
Hi Jerry, Thanks for the comment. Please share what your preference is and why? We’d love to hear!
RE: Compare 7 Best Server…
Dear Mr Sanchez
The topics you’re discussing are of great interest to me. Unfortunately, I do not have a bull’s notion as to what points you’re making, nor any idea of what you’re getting at.
In a way this is my fault, not yours.I would not know a POC from a UCS if either, or both, jumped up and bit me in the leg.
Yet I was directed to this site by asking Google for a “low cost alternative to Dell”. Unlike you, I find Dells dodgy. This may be thanks to their insistence on manufacturing via “dynamic in-transit merge”, a system based on the philosophical precept that it’s good to run an Olympic marathon, and good to build a Swiss watch — and even better to do both at the same time.
So if you, or vminstall, can create a website for The Rest of Us, you might find that there are more of The Rest of Us than you imagine and, not incidentally, far fewer of You Guys than you think. You might, on the other hand, leave that kind of endeavor to those who have no problem communicating with the likes of…
Yours Sincerely,
Paul O’Kelly
My preference is anything dual cpu that can handle lots of memory and is not a true rack server as im also looking for quiet. At work im running lenovo d30s, P700s and P710s as they arw quiet and work well. And they run under my desk in a cube area so quiet is very important.
I have a question…what are your thoughts on rotating servers out? I have 4 Dell PowerEdge R720’s. All four are “listed” as 3 – 5 years EOL. So far no failures or issues outside of replacing hard drives. This is a VMWare 6.0 environment sooo…… Do I ride until they die or start rotating them out? Just looking for a little guidance and another pro’s ideology. Thanks!
Hi Kirk, thanks for the question.
The best practice for server hardware life cycle “in my book” really comes down to your environment, storage, support staff, and budget.
First off, if the real driver for replacing your servers is due to a sysadmin or vendor pushing for the upgrade but there isn’t a real business case such as performance, capacity, or a need for newer OS support then you can probably hold off until a real business case happens.
New servers are great, but often you may need new storage, network gear and costly licenses to go along with them to take advantage of faster, newer and added CPUs.
Then there is the time needed to migrate off the old hardware, which could take weeks or months depending on what ends up getting upgraded along with the servers.
Also, if you do upgrade the servers then you’ll likely upgrade vSphere and ESXI to 6.7, which also adds more complexity and time, and likely additional cost for licenses.
And if you decide to go with more Dells, then you’re looking at spending anywhere between 8 – 12k per node for decent 1U pizza boxes with enough cores and memory to make the upgrade worth doing. as well as give yourself more capacity for growth.
And finally, there’s the time it takes to build the new environment and do the VM migrations off the old cluster that may or may not cause more issues related to SLAs, licenses and hardware dependencies.
All that said, I’m not against upgrading and going through all that I’ve just spelled out. I’m just giving you the big picture because it will take time, people, and a decent budget to do it right.
Another thing to consider might be to get a 3rd party support agreement with a company like Curature or AVT who can warranty the old hardware until you get the new hardware in place.
Kirk, hopefully this helps you decide?
Thanks,
Joe
Thanks Joe!
Really appreciate it! Yeah, that’s kind of where I’m at. Nothing here is so mission critical that if “a host” went down, as long as HA/VMotion works properly, we couldn’t survive. 95% of my file servers simply do nothing more than traditional “file serving.” I only have one actual business app, our ERP system, running in house. 2 if you include Exchange but that, hopefully, will be in the cloud shortly soooo…..
Appreciate the thoughts and insight!
Kirk
That is the perfect post. Great article.
Hi Joe,
I Like your post. Love it. One suggestion you can also include JBS Devices for various brand name server parts. Drives, Memory, CPUs, Power Supplies, Controllers, and other IT Hardware accessories.
You can visit the website to know more.