As you may have read in the news, Microsoft decided to lay off several hundred people from Strategic Missions and Technologies and reorganize the Azure for Operators unit.

I can’t go into any details about what happened internally, but as someone who’s been nine years at the company I can certainly describe what the fallout from this looks like at a personal level:

• I was, miraculously, not impacted (yet).
• Hundreds of people with extremely rare industry expertise (some with over 35 years in telco) are now looking for new jobs across the US, EMEA and APAC.
• Psychologically, everyone who remained is still very much in shock–this was the third round of layoffs we went through in the past few years, and the others were rough, but this one was particularly brutal.

So, to ground myself a bit and put things into perspective, I decided to tally all the previous catastrophes and career-defining moments I “survived”:

• My decision to leave Andersen Consulting (where I was stuck doing SAP and brain-damaging, primitive intranet stuff) to join an upstart ISP that went on to become Novis (and subsumed by what is, today, NOS).
• My decision to leave that ISP and join Telecel (which would later become Vodafone Portugal) and build a few iterations of their ISP services, e-commerce stuff, etc, all of which went through various reorgs and shutdowns.
• Sunsetting Sun hardware in that ISP, which was a seemingly unending political struggle.
• The shutdown of various internal projects as a slow, protracted part of the Vodafone “transition”–which caused no end of fear, uncertainty and doubt.
• The stupefyingly bad decision by Vodafone Group to, in the early 2000s, outsource most of the IT functions to EDS and IBM and move all knowledgeable people to those companies, therefore paralyzing most service launches and causing us years of protracted pain.
• The folly of Vodafone Live! and the insane way in which telcos tried to stem the tide of mobile internet access while racing after the next thing in order to avoid becoming “bit pipes”–which inspired The Blue Packet.
• My first heart surgery, weeks before one of my kids was born.
• The BlackBerry Storm, in which Vodafone first tried to compete with the iPhone, and that I had to do a fair chunk of damage control on.
• The even worse decision to spend billions on Vodafone 360 to, again, compete with the iPhone, which meant I slogged for two years to try to build something I never believed in only to see Samsung sideline the custom hardware they’d built for us and launch the Galaxy brand.
• My move to SAPO and various culture clashes between the Portuguese approach of overpromising and generally winging it and my attempts at a degree of method and practicality.
• The sacking of my entire C-suite as Altice bought Portugal Telecom and began years of quiet corruption that only recently came to light.
• A particularly career-defining moment in my early years at Microsoft (when the transition to Azure and cloud-first was upending the status quo) during which one of my fundamentally incompetent managers told me I would not last the year–I went on to not only prove them wrong but to work on several key projects that paved the way for my move to an EMEA role.
• My second heart surgery, the year before the pandemic.
• And, after some of the best years of my career at Microsoft Consulting in EMEA and the opportunity to follow someone (whom I still very much respect to this day) to Azure for Operators to help structure their professional services, two years of stuff I can’t write about that eventually resulted in what happened this week.

All of these have one thing in common: I had, at best, the illusion of control, but, most often, no ability whatsoever to change the outcome.

And as someone who’s profoundly self-driven and hyper-focused on constantly figuring out the next steps to fix things, being bereft of control and facing a completely unhinged situation (or having agency but being unable to focus it), is the kind of thing that really doesn’t let me sleep at night.

But I’m also writing this to remind me that all of those situations had a before and an after.

And that I shouldn’t really blame myself to be too close to the before to anticipate what could go wrong (well, actually I sort of did, but I can’t write about that…), or (most importantly) forget that there is always an after.

I’m guessing that for the people who have been let go the after currently looks like a blank. Or a turmoil of uncertainty. Or, if you’re the eternally optimistic type (I’m not, but I acknowledge such people statistically have to exist), a bundle of unexplored possibility.

One of the things that keeps going through my mind is that in a job market that is ripe with ageism, where the 5G hype is visibly dying and telcos are left trying to figure out how to recoup the massive spectrum licensing investments, things seem insurmountable¹.

But, again, there is always an after, and that very much depends on finding your sense of purpose.

Even though I technically still have a job, I’m also struggling with that–in all the previous reorgs and shutdowns I went through, finding my purpose again was, unequivocally, the hardest part, so I’m starting small:

• I’m trying to ground myself on “real life” and my hobbies, so that I have a constant reminder that life will go on.
• I’m trying to help those impacted as best I can (wading into the quagmire of LinkedIn “open to work” posts is depressing, but re-posting them or doing referrals to folk I know might help).
• I’m taking calls from other “survivors” and commiserating (of course) but also trying to instill a sense of immediate purpose so that people have some sort of short-term goals while the org reforms–if that means skipping rank and pinging a VP to get some immediate clarity on what we can do, so be it.

And, of course, like any sane person in the technology industry, I am pondering what to do myself. The past few years have demonstrated that job security is no longer a matter of proven technical acumen or delivering consistently good impact, but rather the luck to survive pivots–almost as if big tech was going startup again (with all that implies).

But my key point with this post is the following: the various crisis and setbacks I went through during my career (and life, although I wish I could disentangle them more easily sometimes) taught me that there are always hidden opportunities somewhere, as long as you keep your wits about you and leverage your agency and sense of purpose to rooting them out.

So go out, touch grass, talk to people, form a plan, and get to it.

It’s been a while since I wrote about networking gear, largely because I used to spend a good while working on it and that broke my disclaimer. But home networking being what it is these days, I thought it would be a nice change from my usual fare.

So here’s a quick write-up on the TP-Link L-SG108E gigabit switches I’ve been using for a while now, in the spirit of an ode to the unseen, unheard, unnoticed and unsung heroes of my home network.

Note: This is not a review, since I bought these with my own money and have been using them for a while. Like with the K380, I’m writing this because I’m about to replace them with something else, and I thought it would be nice to write about them before they go.

The Goldilocks Zone for home networking is when all your machines and appliances have great Wi-Fi coverage from a single piece of equipment–or two. And that is what ISPs normally aim for with integrated home gateway/Wi-Fi routers, because most people will be OK with the fiber or DSL connection to be run to a place near their TV (which is where they tend to use their laptops anyway).

In the US, things are even easier because many houses have wooden structures and (ideally) sit somewhat apart, so you can pretty much wing it as far as Wi-Fi coverage is concerned. We’re not far from the days when, thanks to the higher regulated EIRP back in the 802.11b days, you could use a coat hanger and have perfect Wi-Fi signal.

However, things aren’t that simple when, like me, you live in Europe and in a converted flat that is roughly U-shaped, has reinforced concrete walls and wraps around a lift well (or two).

We also don’t have any easy way to route fiber anywhere from the telco box. Vodafone has a Smart Router with a built-in ONT that provides 1Gb/200Mbps to our home, but it has to be mounted in the utility closet and thus has the effective Wi-Fi reach of a beached porpoise (also, telcos tend to have funny ideas about managing your Wi-Fi for you, so mine has its Wi-Fi turned off).

But we have plenty of copper running in the walls–around 25 years ago, as we moved in and tore up everything to run new wiring, I had the foresight of bringing in a spool of honest-to-goodness Lucent Cat 5 cabling and put two jacks in every room, and that was repeated a few years later when we moved some walls around.

And since most of the runs are way shorter than the 100m limit (the flat is weirdly shaped, but not that big), I’ve been able to get away with using that cable for a long time (and have tested 2.5Gb over some of it, so I know it’s good for a while longer).

Yeah, I still think as though I was building mobile networks. Which is good, because our home LAN topology is… complicated, and decent Wi-Fi coverage requires no less than five [Airport Extreme] base stations, all of which are plugged into Gigabit Ethernet.

We have cabling running to several places around the house, some of which have require multiple ports (like the living room, the server closet, and my office), but the resulting topology is not the kind of prototypical star topology I used to design when I put aside that cable spool¹.

Wi-Fi was initially provided by two modded WRT54g on each side of the lift well, but over the years I have progressively upgraded and added to it as we needed better coverage and moved to the 5GHz band (which doesn’t really get past more than one wall)².

But all of it (except for a couple of boutique switches that I use for PoE testing and some higher speed ports) has been connected via TP-Link L-SG108E gigabit switches, and since that’s about to change, I thought I’d write up how they’ve fared so far, since they are (now) dirt cheap and, despite being “unmanaged” in the enterprise way, actually have some management capabilities that I’ve used with decent success.

The TP-Link L-SG108E is a plain (i.e., discreet), run of the mill switch. It has eight Gigabit Ethernet ports, a power port, a tiny reset button, and a few distinguishing features that caught my eye at the time:

• 8 10/100/1000 Mbps GbE ports
• Up to 32 VLANs
• Per port rate limiting and storm control
• Static Link Aggregation
• IGMP snooping
• Port mirroring
• Basic loop prevention

It also supports jumbo frames up to 16KB and has a 16 Gbps switching capacity, but those weren’t top of mind when I got it (and I’ve yet to fiddle with jumbo frames, since I don’t have a pressing need for them).

I was using unmanaged TP-Link switches for a while, and they were fine–but I wanted to try out a managed switch for a few reasons:

• Stability: Given the way our flat was wired up, my network topology is anything but “normal” (again, it’s not really a nice, plain star topology), and even though I have managed to (for instance) provide end-to-end dedicated cable paths between Vodafone’s home gateway to things like set-top-boxes (to essentially get them off my network), I still had issues with IGMP and multicast traffic from IoT devices–including the occasional broadcast storm.
• Traffic Segregation: I wanted to separate my IoT devices and a couple of servers from the rest of my network, and the easiest way to do that is to put them on a separate VLAN. I still can’t get the IoT stuff off the network entirely because the Airport Extremes don’t support VLANs, but my Proxmox machines are using multiple VLANs (with a tiny VM acting as a router).
• Link Aggregation: My Synology DS1019+ has dual Gigabit Ethernet interfaces and it supports link aggregation, so I’ve already tested that–the current disks it has are not particularly fast, but Proxmox backups were much faster, and SMB traffic from my Macs benefited as well.

I’ve reverted it back because I needed to use one of the interfaces for kickstarting a local backup to its predecessor, but I’m planning on setting it up again after the next round of upgrades.

Like I said at the beginning, for a technically “unmanaged” switch the TP-Link L-SG108E has a lot of features, all of which are accessible through a simple, pragmatic web interface:

The first thing I did on all of them was to set up storm protection on the designated “uplink” port, and setting up a few separate VLANs for Proxmox inter-host chatter across two of them, as well as link aggregation on the one in my server closet.

There isn’t really a lot in terms of monitoring other than packet and line error counters, but what there is has already been useful in identifying misbehaving devices (like oddball Realtek NICs). I do like that it has IGMP spoofing and can list multicast groups in use (since I’ve used multicast for my own monitoring for ages).

What I love, however, is that it has port mirroring. I do occasionally need to sniff traffic from devices to figure out what they’re doing, so that’s already been handy a few times–of course I can only do it on the same switch, but more the reason to have a surplus of ports.

Another thing I like about the TP-Link L-SG108E is that it has cable fault detection–I have a cable tester, but when moving things around and reaching into my box of cables to re-patch things I sometimes get either a dud jack (caught one a year ago that was good enough for Fast Ethernet, but had a broken wire for GigE) or a dud cable.

And, sometimes infuriatingly, an intermittent contact, like what you’d get if the cable was physically stressed. Believe me, it helps knowing which cable is “bad” before you start moving heavy furniture around.

I’ve had zero issues setting up 802.1Q on my servers and the TP-Link L-SG108E, or propagating those to another switch. The only issue I had that I can remember was an mis-configuration on my old KVM host (that Proxmox replaced, and provided a nice GUI on top).

I haven’t used the bandwidth “control” feature at all, but the storm control (together with the IGMP snooping) allowed me to limit the damage a buggy IP camera caused sometimes, and what I do is rate limit multicast and broadcast traffic to a point where it’s unlikely to cause damage.

In terms of time saved alone, the three TP-Link L-SG108E I’ve been using (along with their PoE and smaller cousins, which I may write about some day) pretty much paid for themselves even at the EUR 70 price point they originally had (they are now around half that).

They’ve been utterly reliable, quiet, fairly low power and run perfectly cool, so I have no trouble recommending them if you’re either still stuck in the Fast Ethernet age (which is still a reality for most people), or if you need a little more control over your wired network but can’t justify moving to 2.5Gb and beyond.

Like I hinted at in the beginning, the reason I’m writing this is that I will be replacing (some) of them with upgraded gear (which I need now that I have an electronics/hardware test bench and consulting on some industrial projects).

I also want to begin putting in place the right infrastructure to do things like video editing directly on my NAS, which really needs a lot more end-to-end throughput…

That will eventually lead to laying down the foundation for upgrading my Wi-Fi network next year, so… watch this space. There will be more about networking, and likely not all of it about my home.

A few weeks after I posted my review of the K7 Max, Keychron reached out to ask if I wanted to try out the K11 Max, a compact Alice layout keyboard–and I ended up saying yes.

Read More...

This is a tale of ancient wizardry, yearning for control, failed prototypes, and just plain bad timing. Let us begin.

Read More...

A couple of months ago, Deepsounds reached out to ask if I could have a look at their Novation Circuit Tracks patch editor, to which I enthusiastically agreed since I was actually in the process of adding a MIDI IN interface to my MiniDexed and designing printable DIN to TRS adapters so I could use both together.

Read More...

Like I wrote on my review, the TwoTrees SK1 comes with a somewhat serviceable, but quite buggy screen that uses the Nexion UI toolkit. At the time I was already able to use CYD-Klipper to have a remote display, but I’ve been investigating ways to get KlipperScreen working, and finally set up a single-cable, plug-and-play solution:

Read More...