rant

[ceyko]

I skipped a sentence or two but I did not recognize any of that. I wish I did and almost feel like looking it up but it will blow my high. I wana get into IT but it looks like I'm too far off.

Everyone started too far off, a LOT of late nights of work, years of studying, year of experience and then you start writing your own path. IT starts out as crap, then you become a generalist learning everything you can. Then - at least for me - you start specializing in a more focused arena.

i.e. I know little about DBA, but am extremely well versed in all Cisco UC/Telepresence/Route/Switch and moderately (more than most) in most other areas with the Cisco label.

However, I'd be prepared to take a pay hit if you're new to the field. Going into "IT" is like going into "construction" - there are hundreds of different paths/fields/specialties and they change yearly. Figure out what you really like, work to be the best - which will at least result in being better than most that do the work and you're set.

Hell - I think they are nuts - but some people LOVE desktop support. They do it better than 99% of their peers, can communicate their requirements to continue efficient support and as a result it is reflected in their pay check and job role. I do not think I could be paid enough to do desktop support though.

[zora04]

I really hate it when people say sequel

[stang2be]

[GeorgeG.]

[QUOTE=abecx;1548157]I've been doing SQL administration for a long time. I increase performance through query optimization, hardware, configuration settings, data aggregation, storage types, and many other methods that i have learned through practical implementation and reading more books about performance optimization....QUOTE]


damn it... every time you speak, I feel stupid. And I've been in IT for a good bit. But then again, I avoid SQL like the plague.

[GeorgeG.]

I really hate it when people say sequel

LOL, we have this one guy at work that can't pronounce SQL (like sequel). He says sekel so he will literally refer to SQL by saying the letters S, Q, L.

[abecx]

Man if any of you want to have a GTG I can teach you a ton of stuff about hardware level optimization, understanding performance bottlenecks, how the kernel works, and some other low level shit that teaches you how to solve everything honestly. It would be linux based but it carries over to Windows, OsX, Android, basically any operating system using some type of x86 CPU, and even FPGA/ASIC platforms. I can fix anything because I understand the basics so well. I dont shotgun or shattershot troubleshoot, I just use the scientific method, the same shit people have been using for centuries to solve shit but you have to understand the fundamentals to be able to wield it.

I can also teach a SQL class, used to do teach performance at a previous job. I loved/hated it. It would be fun to teach again without it being so corporate rushed, also questions are the beginning of wisdom and if you ask me some shit I dont know, I'll understand the topic(s) even better when we figure it out.

[FunFordCobra]

I know you're not talking to me but if you ever are in Denver, I know all the best dispensaries. Knowledge for knowledge bro.

I wanna learn Linux and start with lower programming like with ras pi for starters.

[jewozzy]

Man if any of you want to have a GTG I can teach you a ton of stuff about hardware level optimization, understanding performance bottlenecks, how the kernel works, and some other low level shit that teaches you how to solve everything honestly. It would be linux based but it carries over to Windows, OsX, Android, basically any operating system using some type of x86 CPU, and even FPGA/ASIC platforms. I can fix anything because I understand the basics so well. I dont shotgun or shattershot troubleshoot, I just use the scientific method, the same shit people have been using for centuries to solve shit but you have to understand the fundamentals to be able to wield it.

I can also teach a SQL class, used to do teach performance at a previous job. I loved/hated it. It would be fun to teach again without it being so corporate rushed, also questions are the beginning of wisdom and if you ask me some shit I dont know, I'll understand the topic(s) even better when we figure it out.

This may be a dumb question but at my work we have to hire a lot of outside people to do stuff to our SQL servers. Is SQL universal? What I mean by that is if you are familiar with SQL could you work on any companies SQL server as far as commands and whatnot?

[abecx]

Yes, SQL is universal. The underlying principals are always the same. Aggregate or index for filesort ( which is cpu based ) performance, remove any massive reference queries and always use the query planner to understand what the platform is thinking.

While the query syntax can change between different platforms, the logic is the same. I've done work on most every SQL type or relational platform out there. I have never had the title of DBA, but I'm highly skilled with it since I have so much experience and have thoroughly read the MySQL performance and MySQL essentials and development books which are almost a line by line debugging of the source code. At this point in my career, when I read a query I already start thinking about how it will retrieve the information, and that knowledge translates to all platforms that are SQL based.

One of my biggest tasks I have done was migrating a platform of 6 billion rows of data with multiple reference/fact tables with 100s of millions of rows from Netezza to MariaDB in order to make a mobile app feasible. The netezza was ridiculously fast for analytics but you put a user on a mobile device and have them wait 15-45 seconds for a response, and they wont be using the app for that long.

I aggregated the data on the netezza since its so fast on an hourly basis and pushed it to a reporting MariaDB server where EVERY query responded in under 0.01 seconds. I saved the launch date and did this in under 3 months with no understanding of the data before hand.

Then to top it off, our API platform was written using a custom ORM, so all of the queries had to be completely redone and they were going to axe the project since we were running out of time and the developers insisted it wasn't possible.

I said with gumption that I could fix it ( I had no idea how at the time ). I ended up using a LUA expression interpreter and converted the netezza queries to MySQL. I only did this because there was a dude that kept pissing me off saying MySQL would never be faster than his $400k netezza platform, and it worked. MySQL has a proxy that does something similar with adjusting queries and I was able to convert the netezza queries to be MySQL compatible using LUA ( something I had never used before either ).

Also, this is how you get your name stuck in peoples minds, by stepping up and taking opportunities. Just hope you can follow through with them, shits paid off drastically for me.

[abecx]

Just some more bullshit, joins are providing different results on one platform versus another. The reason why? The two columns they are joining against have different character sets, aka A != A in latin1 vs utf8. It worked on MySQL because it just doesnt care, MariaDB enforces it. It also slows their query times drastically because it has to convert the character set every time the query is run and prevents using indexes.

[Craizie]

I've understood maybe four words you've written, but I'd be down for a get together on this shit. I've been looking to make a career change for a while.

[abecx]

Just some more bullshit, joins are providing different results on one platform versus another. The reason why? The two columns they are joining against have different character sets, aka A != A in latin1 vs utf8. It worked on MySQL because it just doesnt care, MariaDB enforces it. It also slows their query times drastically because it has to convert the character set every time the query is run and prevents using indexes.

One of the hardest things to do is admit you're wrong. I have no problem admitting I am wrong because it was at the time made with available evidence.

The issue turned out to be after more investigation, not the character sets or the indexes, it turned out to be the query.

MySQL engine has a query optimizer which generates a plan on gathering your results and displaying them, when I was adding specific charset or ignoring indexes to the query, I would get different results. The reason why is the query had an aggregate group by statement that did not cover all the select fields. This means that MySQL's query planner can show different results for the same query based on outside variables. Fucking mouthful, but basically the query is not compatible with all versions of MySQL because the engine changes between versions ( it only works on 5.5.37 ).

People pay for this level of research. JUST RUN THE SAME FUCKING VERSIONS FOREVER is never a valid response as much as developers may hate it, it is very important to know where your root cause is because developing reports that are accurate is vital.

[Nash B.]

One of the hardest things to do is admit you're wrong. I have no problem admitting I am wrong because it was at the time made with available evidence.

The issue turned out to be after more investigation, not the character sets or the indexes, it turned out to be the query.

MySQL engine has a query optimizer which generates a plan on gathering your results and displaying them, when I was adding specific charset or ignoring indexes to the query, I would get different results. The reason why is the query had an aggregate group by statement that did not cover all the select fields. This means that MySQL's query planner can show different results for the same query based on outside variables. Fucking mouthful, but basically the query is not compatible with all versions of MySQL because the engine changes between versions ( it only works on 5.5.37 ).

Noob mistake.