Tidbits, July 2023

A few tidbits on my mind at the moment, including:

• Can you ever find the perfect tool?
• The inverse-square law for recovery time objectives
• Raging against the (smart) machine
• Scamception
• Coffee, and
• Retro computing.

Let’s get started.

The endless pursuit of the perfect tool

I’ve been happily using Notion for my personal note taking for the last 6+ months, having ‘settled on it from ‘migrated’ across from OneNote. (I still, lamentably, need to use OneNote for work.)

Every now and then I have to sit and remind myself of the tyranny (or paradox) of choice, and recently I bumped up against this when someone introduced me to Obsidian. It looks like a really cool note taking tool and certainly has some advantages over Notion. So I found myself toying with the idea of ditching Notion for Obsidian. I ended up downloading Obsidian on a Windows laptop and my Mac to spin it up. Like Notion, there’s a learning curve associated with it, but there’s also some cool features for people who take a lot of notes and … I found myself forgetting the basic rule of productivity: nothing is perfect, so once you’ve made your choice, and unless it’s a terrible choice, you’re better off sticking with it.

The tyranny/paradox of choice goes like this:

People are given too many choices and this leads to them feeling less satisfied with their decision. This often happens when people are bombarded with choices and cannot narrow down their options.

“What Is the Illusion of Choice?”, Arlin Cuncic, April 28 2022, VeryWellMind.

Whenever there’s more than one tool available for the job, there’s a high probability that none of those tools will be “chef’s kiss” perfect for the role. There’ll be compromises you have to make. But unless they’re egregiously terrible compromises (e.g., OneNote), or your needs have seismically shifted, once you’ve made that choice you’re probably going to stay more productive by sticking with the tool you chose rather than jumping onto the next tool that comes along. You could also call this the “shiny new toy syndrome”.

Case in point: over a decade ago I lurched between a succession of TODO apps, trying to find the right tool to help me stay organised with my work. Eventually I came to the conclusion that they were all crap — not because of what they did, but because they didn’t suit how I worked. For me, as soon as the TODO app is closed, I lose context and forget to do things. So my tool is decidedly old-school: a paper-based journal, one page per day, writing out all the things I need to do and the meetings I have to attend, and I can tick them off. And because that’s a book that stays open beside my keyboard all the time, I stay substantially on track (or as much as random interrupts to my schedule allow). Perfect tool? No. Search is terrible. No copy and paste. No electronic dictation. No pop-up reminders. But it works and so it’s not worth pursuing non-existent perfection elsewhere. (And it lets me use fountain pens on a daily basis.)

The inverse-square law of RTO

Imagine a common backup schedule for a database, such as the following:

• Log backups every fifteen minutes, kept for 7 days
• Daily* backups, kept for 31 days
• Monthly backups, kept for 12 months
• Yearly backups, kept for 7 years

There’s a general rule of thumb that backups you need to use for operational recoveries need a tighter recovery time objective (RTO) than backups you’re keeping for compliance or long-term retention recoveries — and this becomes increasingly strict as the criticality of the workload increases to the business. Being a Doctor Who fan I sometimes joke that the ideal technology for meeting the operational RTO for a mission-critical database involves a TARDIS.

As the age of the backup increases though, not only does the recovery point objective (RPO) shift, but perhaps more importantly — the RTO expands. The above set of schedules might be mapped into RTOs as follows:

When backup was taken	Example RTO	Likely Purpose of Recovery
Within the last day	At most: As long as it took to do the backup. Preference: Faster even than that.	Disaster recovery, recovery from corruption, etc.
24 Hours – ~3 days	Ideally no more than 110% of the time it took to do the backup. It needs to be fast, but there’s an increasing focus on whether the recovery speed might impact the network or other systems.	Disaster recovery, recovery from corruption, etc.
4 days – 2 weeks	110% – 200% of backup time. Recovery can definitely take longer than it took to backup, particularly since there’s a high chance it’ll go to another system with fewer resources.	Data refresh, Recovery Auditing
Rest of operational retention window.	200%+ of backup time. After about two weeks is where some businesses start to think about off-loading even their operational retention backups to slower storage. While the economics of the offload don’t always work here, the focus has clearly shifted from high speed recovery to cost-efficient recovery.	Data refresh, Recovery Auditing, other test use cases
Medium-term compliance retention	Often 2-7 days. (The 12 months retention schedule I mentioned.) At the point of it being a recovery from compliance retention, there’s a recognition that it may take longer to recover, mainly because of the storage medium that it’s sitting on. Maybe it got moved out to object, and it’s going to take a while for it to come back from that slower storage.	Recovery auditing, Compliance retrieval
Long-term compliance retention	Often 7+ days. Sometimes as long as a month. Once we’re recovering from long-term compliance retention, we’re not only dealing with slower storage (e.g., object), but also potential compatibility creep. Maybe it was a backup of an Oracle 18c database, and all the database servers now are running Oracle 23c. So recovery isn’t just about getting the data back from slower storage, it’s actually about spinning up a usable recovery environment.	Recovery auditing (less frequently), Compliance retrieval

Of course, everyone wants a fast RTO and any time you can deliver a faster RTO is a win. But achieving high speed RTOs across the entire lifecycle of a backup gets increasingly challenging and expensive, so practicality comes into play.

Is it a true inverse-square law? Not exactly, but it is a good analogy to use, particularly when trying to explain to anyone how RTOs shift over time.

(*Daily: Note I don’t talk about weekly here. I’m not a big fan of weekly backups unless you’re doing something particularly special. Otherwise your end of week backup is just another day of the week.)

Rage against the (Smart) Machine

There’s a meme that goes around regularly along the lines of “Rage Against the Machine never said what machine they were raging against, but I’m pretty sure it was a printer”.

Printers regularly fall into the realm of rage-inducing user-experience, but I’m going to go out on a limb and suggest that if Rage Against the Machine were forming now, they’d be looking squarely at so-called Smart Appliances.

I’ve got three so-called Smart Appliances in my house at the moment:

• A TCL LCD TV
• A Miele washing machine
• A Breville panel heater

The TCL TV became so interruptive to our viewing process by wanting to throw up information alerts to screen about random Android component updates (rather than say, just doing them between the time we switch the TV off and when it goes off) and discovering ‘new’ channels (they never were) that the only solution was to turn its WiFi off and make sure it was disconnected from ethernet. No network connection, no annoyances. (Yes, that may mean no security updates, but with it being disconnected from the network I’m willing to take that risk.)

The Miele washing machine (in addition to feeling the need to incessantly beep for 10 minutes after a load is finished) had no option whatsoever for initialising without getting connected to the Internet. Sure, it has an app that tells you when you’re out if your washing is finished, but if you don’t consider that to be urgent information, you’re just plain out of luck — Nanna needs a Miele? Well, Nanna needs WiFi.

And the Breville. Its app is apparently so ‘smart’ that it can warn me, when I do the initial setup for it, that I have a space in my WiFi password (this, apparently, warrants warning). Why do I recall something that’s only needed on initial setup? Well, that’s because I’ve done the initial setup now 20+ times since I bought it — and for a very simple reason: every time our ISP flips our IP address, Breville’s systems decides the heater has been permanently disconnected, requiring it to be removed from the app and configured from the start again.

Look, I’m prepared to say that I could have just randomly lucked out, three for three, with Smart Appliance purchases — but I think the more likely scenario (given what I see in chatter online about Smart Appliances) is that their user experience sucks. I’m a highly technical Gen-Xer with a degree in computer science and these things bamboozle me when they fail. If I look at these products with the lens of my 75 year old mother who thinks “oh well I guess it just doesn’t work any more” when Netflix forgets her password, these Smart Appliances are not.

And that’s the problem. In product management we talk about ‘happy paths’ and ‘unhappy paths’; the former is about when things go right, and the latter is about making sure you account for situations where things go wrong. Of course, you want to make sure that the happy path experience is good, but that’s nothing compared to needing to make sure that the unhappy path experience is fantastic. I.e., don’t make a potentially undesirable situation worse by having a crappy way of handling predictable situations like:

• An update comes in while someone is watching the TV,
• A person who doesn’t have WiFi but bought a new washing machine that has some user experiences available requiring it,
• The heater occasionally loses power or internet connectivity.

If companies put even half as much effort into making their Smart appliances Civil, we’d actually have a decent ecosystem. Until then, I’m going to be the curmudgeonly old guy who does my best to either buy dumb appliances, or disable the so-called ‘smarts’.

Scamception

I carry two mobile phones — one for work, one for home. That has its advantages, but one area where I definitely pay the price is the volume of scam calls I get. Particularly on my personal phone (friends & family know to text me), if I get a call from an unknown number it’s going to be:

<Robot voice>: Hello, this is {eBay | randomBank | Australian Taxation Department | Australian Federal Police | Amazon}

I was surprised yesterday to get a scam call that instead started with:

<Robot voice>: Hello. Are you tired of scam calls? Our scam call blocking service will...

I guess they know their market. Everyone is sick of scam calls, so maybe someone (sorry, looking a redacted age demographic) will frantically mash the 1 button to talk to someone to fix their scam call problem. But I guess it had to happen — scam calls are so prevalent now that getting a scam call that purports to stop scam calls is the new getting an email from NacAfee AntiVirough (urgentpaynow@yourantivirussoftwarehasexpired.wearetotalylegit.ru).

Pod people

My odd working hours (2.30am alarm set three days a week on average) mean I need to be self-sufficient in my office space to avoid disturbing my husband. One of the ways I achieve that is to have a pod-based coffee machine near my desk — it’s quieter and more accessible than our fully automatic espresso machine in the kitchen. But of course, it’s hampered by the fact that almost all pod coffee is vile. Nespresso’s pods, Starbuck’s pods, all the major brands seem to work on the basis of “quality coffee is for losers”.

One of the cool things about having a husband who works in the magazine industry is that he happens to work on a couple of coffee magazines, and a month or so ago when we were in the supermarket he noted that amongst the coffee pod section there was a premium Melbourne brand that he sees a lot in his magazines. On spec we bought a couple of options (“Wide Awake” and “Orthodox”) and oh my.

So, while I still prefer coffee in espresso format made straight from the on-demand ground bean, I can heartily recommend St Ali’s coffee pods. (It’s just a link — I’m not getting any kickback, but I always get a kick from recommending good coffee.)

Excuse me, I’m a Mummified Husk

One of the things I do enjoy about social media is getting to interact with people who are substantially younger than me. There’s nothing like talking to someone in their twenties to give me (soon to turn fifty) a different perspective. But, then I saw someone post recently:

“I’m into retro computers and games — ones from the early 2000’s.”

Well, that did make me feel old.

So, this withered husk of pre-retro computing will leave you with the greatest sound track ever written to an 8-bit computer game, from 1985. In fact, at 5+ minutes, you’d argue that for the C64 version, you paid for an 8-bit synth classic and got a complimentary game thrown in.