The Register® — Biting the hand that feeds IT

Feeds

Mmmm, TOE jam: Trev shoves Intel's NICs in his bonkers test lab

If you want to impress me, make kit that 'just works'

SaaS data loss: The problem you didn’t know you had

Review Tech offerings are rarely "good" to me, merely "less awful than others." In this case, however, I'll be looking at the complete package of Intel networking's efforts - the hardware, software and ongoing support - and might even attribute that word to its efforts. This isn't to say that everything in the Intel networking universe is awesome - I've certainly encountered some weird stuff over the years with its cards. In the majority of cases, however, the answers were to be found on Intel's own support website rather than as esoteric knowledge of the sysadmin mafiosi.

Intel's networking folks serve as stark contrast to my experiences with Nvidia's networking components. There is a special hell reserved for telemarketers and whoever designed the nForce NIC/forcedeth driver combo. I consider it a professional failing as a writer that I cannot create a sufficiently vitriolic phrase to encompass the depth of my discontent with that product. I simply lack the words.

But when it comes to Intel, not only were workarounds or potential fixes posted, but fairly upfront explanations as to why the bad thing was happening were generally posted as well.

Intel in VMware (click to enlarge)

If the fix needed to come from Intel's end, you stood a good chance of getting its networking guys to issue a fix. At the very least, it seemed willing to work with the open source community to make sure that – if we wanted to – we could build a fix on our own.

The end result was hardware that would "just work". When it came to some of the older Gig-E NICs, I would be lying if I said that Intel's offering delivered the highest speed or the lowest latency. Other vendors offered cards with more features, more offloading, and ultimately better performance... if you had everything set up just right.

But those setups were fragile, and they certainly weren't something I was prepared to use in deployments where I knew that the company would end up repurposing that server for some other task down the line. Many a Windows-based database server has ended up a VMware host in the past few years. For me, network card compatibility has proved the number one issue with those conversions.

The old and the new

While I've got a lab full of Intel 1Gbit cards that seem to work in everything, the truth of the matter is most of this stuff is old. Half my inventory consists of Pro/1000 cards that have been in service for over a decade. Until very recently my only 10GbE Intel cards were a pair of XF SR Dual Port Server Adapters, and those are from 2007.

Network Connections

As luck would have it, Intel was convinced to part with some newer cards for testing. I have a pair of X520-DA2s two I250-T4s as well as a pair of X540-T2s. In essence, I have managed to assemble "the evolution of Intel Ethernet networking" for the past decade or so.

Steps to Take Before Choosing a Business Continuity Partner

Next page: Convergence

Re: most of this stuff is old

Useless?

4
0

Any chance you can get their networking software guys to have a quiet word with their graphics software guys about how to write a driver that actually vaguely works?

3
0

Re: Terrible review

And you're wrong. Any network card for which you have good, low-level access can be reconfigured to send non-standard Ethernet frames. It takes a little bit of bit-bashing on the driver creator's part, but you can turn a regular old network card into something that will make FCoE frames.

What you cannot do is send those frames over a standard ethernet network unless you have similarly updated your switch. To be clear: you are not going to be getting firmware from D-link to this, but you can usually get your higher-end Cisco stuff upgraded to handle the non-standard frames. This means that you can do FCoE point-to-point only unless you invest in the right infrastructure (which should include CNAs, make no mistake) but that you can make a NIC speak FCoE frames if you tinker with it enough. (Nobody does it because what would be the point?)

The fact that you've never rewritten a firmware or driver (or done any real bit-banging) doesn't mean others haven't. Please bear that in mind the next time you wander around accusing people of things.

You'll also note that while I said that a regular network card could be made to speak either protocol, I only discussed iSCSI as being in in sort of practical use without a CNA. And now we've had this little conversation in the comments so there is even more information available. Internet!

2
0

More from The Register

SCO vs. IBM battle resumes over ownership of Unix
Zombie lawsuit back and wants to suck the brains out of Linux
 breaking news
You don't need phone lines or cable for ANYTHING, says Dish
The satellite-dish man can sort you out with phone and broadband over the air too
 breaking news
What's HP got under wraps? Looks awfully flash and tape shaped
What happens in Vegas won't stay there - we've got the details
Microsoft borks botnet takedown in Citadel snafu
Stupid Redmond kicked over our honeypots, wail white hats
IBM's $1bn layoffs latest: Now axe swings in US, Canada - reports
Union claims 121 storage bods canned after dismal sales
NetApp musters muscular cluster bluster for ONTAP busters
Storage array OS overhauled to juggle more nodes, go down on you, er, less
HP adds 'Haswell' Xeon E3s to entry ProLiant servers
Gussies up MicroServer for SMBs, adds baby switches