Data Leak, Work Trips, Concurrency, ERP, OVH, HA, Thunderbird, Contrast
Post date: Apr 30, 2017 5:55:34 AM
- I think I lost one USB stick. Well, doesn't really matter. Everything is always encrypted and backed up, so it isn't that big fuss. But some people just don't do it right. - After all that USB stick was recovered. It was just bit incorrectly located and therefore outside the expected search and recovery perimeter. Good thing was that after recovery it was easy to notice that yes, version control was up to date and data was encrypted, just as designed. Practically losing it would not have made any practical difference, just monetary loss of around 20€ or something. But naturally things could have been much worse.
- Sometimes its funny when people are jealous about work trips. Don't be, there's a reason why those are called WORK trips. It just means bunch of very long days and hanging in taxis, airport and planes.
- 15 years of concurrency - Very nice blog post about the issues I've been covering on my blog over and over again. Because those things still cause trouble today. kw: Software Transactional Memory (STM), Threads, Locks, Multicore, Manycore, Atomic, Transactions, Transactional Synchronization Extension (TSX), Parallelism, Parallel, MapReduce, Shared nothing, Event-loop concurrency. Sequential Consistency, Tear-Free Code.
- Read a few common ERP mistakes and how to avoid making them articles. All of that was just so normal. Nothing new. But that doesn't surprise me. I've got 20 years of experience in ICT projects.
- Studied OVH Public and Private Cloud offerings in detail. Even more discussion about processes, threading, locking, performance. Per core speed, overall number of cores. Amount of RAM, RAM speed, storage system speed, tiered storage and experienced performance vs measured performance. And so on. Well well, nothing new with this. Should systems be business critical, what's the SLA level, what if SLA isn't met. How quickly support should react and other 'business as usual' cloud / outsourcing discussion. With multiple service providers as well as with technical software development guys, what are the actually requirements and what kind of hardware setup is the best for our use cases and so on.
- Some pretty sarcastic discussions about High Availability (HA). Because as it seems, the High Availability is actually often the reason why systems aren't available. Much simpler setup without HA would have worked just fine. But when complex and cool HA stuff is being used, then everything starts to fail really miserably. - This is a good example, we have this ultimate super complex bs bs bs technology. When you start using it, you'll get exactly the opposite what you were expecting to get. So using high availability caused a lot of downtime, isn't that funny? In a way yes, but as stakeholder I don't think it's funny at all. - It's like a fighter jet with randomly firing ejection seat which leads to destruction of the jet and death of the pilot. Cool, we got this life saving cool high tech device, which actually just downed a jet and killed the pilot. That's awesome, not.
- Curios case of data loss with Mozilla Thunderbird. When moving messages from IMAP folder to Local Personal Folders Thunderbird can crash at anytime. When it crashes, it has already removed messages from IMAP storage, but only created index entries in local message folder. Actually all message data is missing. This is yet another great example, how hard even simple computer things are for engineers. Making message moving from folder A to B reliably seems to be really hard problem. If you've seen all the whining about similar issues and sucky transactionality and stupid software architecture and design, you'll know that this isn't anything new. Actually this is just the usual case. I just which this would be exception and not the norm. But that's just a pretty unicorn pooping rainbows dream from meme fantasy land. If you check out Bugzilla@Mozilla you'll find out that they've been fixing this issue for over 10 years!
- About those silly writings earlier and contrast. Now one guy has been building led light with 5 kW power. This is starting to be on the lines I wrote earlier. I wonder when display can deliver about 120k lux light intensity than mid day sun, then it's starts to be enough. Btw. You might need 20 - 30 kW of power for that display, but at least it can deliver enough contrast and brightness. Remember to wear welding glasses when looking at screen, so your eyes won't get damaged. So much about that bigger and more is better. Good enough, must be good enough. More isn't always better. Of course if you're playing CS it's nice if you can actually feel the effects of flash bangs, instead of only getting software simulated version.