Why I Upgraded srv02 from Xeon X5560s to X5690s
One of the recurring themes in this homelab is that I like old enterprise hardware right up until it gets in the way.
That's where srv02 ended up.
It's a Dell R710, which means it's from that nice era of servers where everything feels overbuilt, replacement parts are easy to find, and the machine is just new enough to still be useful if you're realistic about what it is. Mine had dual Intel Xeon X5560s, which were perfectly serviceable for general BSD workloads, storage-adjacent tasks, and the usual "this should live on a separate box" experiments.
But I kept running into the same wall: virtualization.
The Actual Problem
The issue wasn't cores. It wasn't clock speed. It wasn't even memory.
It was CPU feature support.
The X5560s have Intel VT-x, which sounds like it should be enough until you discover that "has virtualization" and "has the virtualization features modern hypervisors actually need" are not the same thing. What I needed on this machine was unrestricted guest support. The old CPUs didn't have it.
That meant srv02 sat in an awkward middle ground: powerful enough to make me want to use it as a VM host, but missing the specific VMX capability that makes that practical for bhyve or other hardware-assisted virtualization workloads.
I don't mind software limitations. I mind hardware limitations disguised as "almost."
Why the X5690s
The fix was straightforward: swap the X5560s for dual Intel Xeon X5690s.
They're still from the same general platform generation, still fit the machine cleanly, and still make sense in an R710 without turning the whole thing into a science project. More importantly, they add the virtualization features the box was missing — including unrestricted guest support.
That changes the machine from "maybe useful for virtualization if I squint" to "actually capable of doing the thing I bought it to do."
As a bonus, the X5690s are just better CPUs overall. Higher clocks, more cores, more threads. But that's not really why I did it. If all I wanted was a modest performance bump, I could have kept putting it off.
This upgrade was about capability, not benchmark vanity.
Why I Didn't Just Replace the Server
Because the rest of the machine was fine.
This is the part that always gets lost in homelab discussions. The answer to every hardware limitation doesn't have to be "buy a newer server." Sometimes the right move is a targeted upgrade that extends the useful life of the hardware you already have.
The R710 still has plenty going for it:
- cheap parts
- known behavior
- enough expansion for what I need
- no mystery firmware story
- no need to redesign the rack around it
Swapping CPUs is a lot cheaper, and a lot less annoying, than replacing a whole platform.
The Upgrade Itself
This was one of those upgrades that feels bigger in theory than it is in practice.
Open the box. Pull the heatsinks. Swap the processors. Re-seat everything. Re-apply thermal paste. Put it back together. Boot, then immediately stare at the POST screen longer than necessary because that's what server upgrades do to you.
The R710 took the X5690s without drama, which is about the best thing you can say about a hardware upgrade. No strange compatibility issues. No unexpected firmware detours. Just the usual amount of mild anxiety followed by a successful boot.
That's my favorite kind of maintenance: the kind that ends with the machine acting like nothing happened, except now it can do more.
What Actually Changed
The practical outcome is simple: srv02 can now be used for the virtualization workloads I had in mind when I started looking at bhyve and similar tooling.
Before, the machine technically supported virtualization in the vague, marketing-bullet sense. Now it supports it in the way that matters.
That's an important distinction, especially on older hardware. A lot of secondhand servers sit in this exact category: they look like obvious VM hosts until you check the processor feature matrix and realize you're one generation short of what you need.
This is one of those deeply unglamorous homelab lessons: always check CPU flags before you design around a machine.
The Part I Found Most Annoying
What bothered me wasn't that the X5560s were old.
It was that they were close.
If they had been obviously unusable, I would have replaced them immediately and moved on. But they were competent enough everywhere else that they kept encouraging the idea that I could make them work for one more use case. That's how you end up spending hours reading virtualization documentation, forum threads, and CPU spec sheets for processors released more than a decade ago.
Sometimes old hardware saves you money. Sometimes it charges you in research.
The Honest Takeaway
I didn't upgrade srv02 because I wanted faster CPUs.
I upgraded it because I wanted the machine to stop being almost capable.
The X5560s were fine right up until the moment I needed a specific hardware feature they didn't have. The X5690s solved that cleanly, preserved the rest of the platform, and let the R710 keep doing useful work instead of becoming another retired box I swear I'll repurpose later.
That's usually the real goal in a homelab: not chasing the newest thing, just removing the one bottleneck that's preventing a good machine from being the machine you actually need.