Not all cloud instances are created equal

You know how anecdotally we say, “in the cloud, bad nodes exist” so you should always get a baseline?

Today I ran a query (repeatedly) on sqlite3, and on two instances, I got time measured as:

Run Time: real 13.405 user 13.243332 sys 0.046667
Run Time: real 10.989 user 10.963332 sys 0.010000

This naturally skewed results I initially got for something related to MySQL/MariaDB benchmarking. It really was such that while I had 2 instances, in the same region/AZ, I had one “good” node and one “bad” node.

Updating the locate database on macOS

Today I uninstalled CrashPlan on my macOS laptop since on August 22 2017 they announced they were exiting the home user market. I think the party is close to being over as even Amazon CloudDrive is now limited to 1TB of storage.

I wanted to ensure I removed any traces of CrashPlan. I use the locate command a lot. Ever wondered how to run the equivalent of updatedb on macOS? Its locate.updatedb (comes with a man page too). Run sudo /usr/libexec/locate.updatedb and that updates the locate database.

Pleased to say CrashPlan went away quite cleanly.

Keybase and FUSE on macOS

I like the concept of Keybase, have had an account there for sometime (https://keybase.io/bytebot). I also installed the Keybase app on macOS which has kbfs, the FUSE-based filesystem for Keybase users to share files with encryption (read more).

However, I also tend to spend time offline, and sometimes on restricted networks where I use TripMode. I realised very quickly that a simple ls or df could take some 20 seconds to complete (timeouts at work).

After poking around I realised it was Keybase. So this is what I do now:

keybase ctl stop
pkill Keybase

I used to have to also unmount the /keybase directory but this seems to be taken care of by newer clients. A macOS annoyance. I can’t always just have it running. There is a github issue #971. I think their iOS app is quite well done and have no complaints thus far.

Writing on the Percona blog, upcoming appearances

I’ve started a column at the Percona Database Performance Blog, and some recent posts (I don’t plan to record all this; I might just upload the pre-edited Markdown to Github at some stage):

Watch that blog every Friday.

I also have some upcoming appearances and I look forward to meeting you there:

Grab bagged South East Asia? Uber still in the running

I see a lot of anecdotal evidence that people claim Grab has South East Asia in the bag and Uber is doomed. I beg to differ.

I take a Grab or Uber from KLIA quite a bit. I always try the Grab app first. Take yesterday (Friday night/Saturday morning). I landed at 1am and tried the Grab app twice: one for GrabCar and another for JustGrab. It found no rides.

Contrast that to Uber. Tried for UberX once and I was in a car within 5 minutes. This is generally my experience when it comes to using Uber versus Grab at the airport.

I try to speak to drivers for a bit before delving into reading on my tablet. Today my driver whom has completed 840 trips on the Uber platform in the 9 months he has been on it told me he is Uber-exclusive. He knows Grab exists but he’s all in on Uber.

Some drivers do drive for both platforms. But it is clearly not all.

So my experience so far has been that I’m doing better with Uber even though I try Grab first. And I estimate my data points from KLIA are approximately 3 rides per month.

Of course, not having to use Alm Cabs “airport limousine” is clearly the bonus here!

Vertu no more

Vertu has wound up.

Was this a case of a string of management failures? This company started at Nokia [sic] in 1998, then sold to a private equity firm EQT in 2012 for more than €200m, then to an obscure Chinese company called Godin Holdings in 2015, then in March 2017 to Murat Hakan Uzan, a Turkish exile based in Paris.

I have only ever seen one Vertu phone in the wild (on a Cathay Pacific flight in business class; she had a 12.9″ iPad Pro as well). I did see a couple of their stores still open and very much trying to flog their wares (Oriental Plaza Beijing, amongst a few others; though I suspect they were not owned by Vertu directly or it takes some time for international foreclosure to work).

FT’s Jonathan Margolis (“UK’s original consumer tech journalist”) sings praises for Vertu. In 2012 he wrote an official company history that was never published as the CEO who commissioned it, departed (warning bells should have already gone off then). Vertu apparently did well under Nokia’s stewardship, and even a year into their new owners; in total they sold 500,000 of their phones.

What struck me as well is how many “luxury” companies bet on China. What China giveth, it can also taketh: “Restrictions on business gift-giving in China from 2014 onwards affected Vertu. The phones were a popular status symbol there. One small city, Wenzhou in Zhejiang province, had two Vertu stores.”

I will respect this one brand aspect: Vertu in its pre-2013 heyday never gave away phones to celebrities. Why don’t more brands realise this? Consumers are getting smarter about product placements.

They apparently also had a niche userbase: Vertu phones from 2002 are said to still work. Updated versions of early voice-and-text-only models were its best-sellers until the end. Many Vertu owners used phones only for calls.

All in, its sad to see a brand I’ve known for so long, reach its demise. It seems like if you owned a Vertu, you were really part of an exclusive club, which is why I never saw many in the wild.


i